demand_forecasting_in_supply_chain_analytics.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/malikbaqi12/Applied-data-science-using-pyspark-code-files/blob/main/demand_forecasting_in_supply_chain_analytics.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "60c5172f"
      },
      "source": [
        "## Demand Forecasting with using Big Data"
      ],
      "id": "60c5172f"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5a6e642b"
      },
      "source": [
        "## Content\n",
        "1. [Introduction](#section-intro)\n",
        "2. [Importing libraries and Kaggle setup](#section-ts)\n",
        "3. [Load Dataset](#section-pro)\n",
        "4. [Basic Exploratory Data Analysis](#section-ten)\n",
        "5. [Feature Engineering](#section-ten)\n",
        "6. [Data Encoding](#section-ten)\n",
        "7. [LightGB Model](#section-ten)\n",
        "8. [Time Series Analysis](#section-ten)   "
      ],
      "id": "5a6e642b"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "568a11d6"
      },
      "source": [
        "## 1. Introduction"
      ],
      "id": "568a11d6"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5acc8a6e"
      },
      "source": [
        "#### Dataset Overview\n",
        "* A store chain's 5-year data includes information on 10 different stores and 50 different products.\n",
        "* The data set covers the period between 01-01-2013 and 31-12-2017."
      ],
      "id": "5acc8a6e"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "f510ab31"
      },
      "source": [
        "#### Business Problem\n",
        "* It is desired to create a 3-month demand forecasting model for 10 different stores and 50 different products of a chain of stores.\n",
        "* Afterwards, it is desired to reduce the data set to weekly and create a demand forecasting model for 2017."
      ],
      "id": "f510ab31"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "951e2930"
      },
      "source": [
        "#### Variables\n",
        "* date – Date of sales data (No holiday effects or store closures)\n",
        "* store - Store ID Unique number for each store.\n",
        "* item - Item ID Unique number for each item.\n",
        "* sales – Number of items sold, Number of items sold from a particular store on a given date"
      ],
      "id": "951e2930"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9d172d27"
      },
      "source": [
        "## 2. Importing Libraries and Kaggle Setup"
      ],
      "id": "9d172d27"
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "id": "5e81be15"
      },
      "outputs": [],
      "source": [
        "# Importing Libraries\n",
        "\n",
        "import numpy as np \n",
        "import pandas as pd \n",
        "from matplotlib import pyplot as plt\n",
        "import seaborn as sns\n",
        "import lightgbm as lgb\n",
        "from statsmodels.tsa.holtwinters import SimpleExpSmoothing\n",
        "from sklearn.metrics import mean_absolute_error\n",
        "from statsmodels.tsa.holtwinters import ExponentialSmoothing\n",
        "from statsmodels.tsa.statespace.sarimax import SARIMAX\n",
        "from statsmodels.tsa.arima_model import ARIMA\n",
        "from statsmodels.tsa.seasonal import seasonal_decompose\n",
        "import statsmodels.api as sm\n",
        "import itertools\n",
        "\n",
        "import warnings\n",
        "pd.set_option('display.max_columns', None)\n",
        "pd.set_option('display.width', 500)\n",
        "warnings.filterwarnings('ignore')\n",
        "\n",
        "#Kaggle setup\n",
        "import os\n",
        "for dirname, _, filenames in os.walk('/input'):\n",
        "    for filename in filenames:\n",
        "        print(os.path.join(dirname, filename))\n",
        "\n"
      ],
      "id": "5e81be15"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "13bc2b43"
      },
      "source": [
        "## 3. Load Dataset"
      ],
      "id": "13bc2b43"
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 520
        },
        "id": "090b71f6",
        "outputId": "1a50d6fe-dbf0-46d8-9c88-e729648b2072"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "         date  store  item  sales  id\n",
              "0  2013-01-01    1.0   1.0   13.0 NaN\n",
              "1  2013-01-02    1.0   1.0   11.0 NaN\n",
              "2  2013-01-03    1.0   1.0   14.0 NaN\n",
              "3  2013-01-04    1.0   1.0   13.0 NaN\n",
              "4  2013-01-05    1.0   1.0   10.0 NaN\n",
              "5  2013-01-06    1.0   1.0   12.0 NaN\n",
              "6  2013-01-07    1.0   1.0   10.0 NaN\n",
              "7  2013-01-08    1.0   1.0    9.0 NaN\n",
              "8  2013-01-09    1.0   1.0   12.0 NaN\n",
              "9  2013-01-10    1.0   1.0    9.0 NaN\n",
              "10 2013-01-11    1.0   1.0    9.0 NaN\n",
              "11 2013-01-12    1.0   1.0    7.0 NaN\n",
              "12 2013-01-13    1.0   1.0   10.0 NaN\n",
              "13 2013-01-14    1.0   1.0   12.0 NaN\n",
              "14 2013-01-15    1.0   1.0    5.0 NaN"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-82255a24-269d-4c6f-953d-4f9a4d3e6998\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>date</th>\n",
              "      <th>store</th>\n",
              "      <th>item</th>\n",
              "      <th>sales</th>\n",
              "      <th>id</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>2013-01-01</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>13.0</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>2013-01-02</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>11.0</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2013-01-03</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>14.0</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>2013-01-04</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>13.0</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>2013-01-05</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>10.0</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>2013-01-06</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>12.0</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>2013-01-07</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>10.0</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>2013-01-08</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>9.0</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>2013-01-09</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>12.0</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>2013-01-10</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>9.0</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10</th>\n",
              "      <td>2013-01-11</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>9.0</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>11</th>\n",
              "      <td>2013-01-12</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>7.0</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12</th>\n",
              "      <td>2013-01-13</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>10.0</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13</th>\n",
              "      <td>2013-01-14</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>12.0</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14</th>\n",
              "      <td>2013-01-15</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>5.0</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-82255a24-269d-4c6f-953d-4f9a4d3e6998')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-82255a24-269d-4c6f-953d-4f9a4d3e6998 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-82255a24-269d-4c6f-953d-4f9a4d3e6998');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 2
        }
      ],
      "source": [
        "train = pd.read_csv('train.csv', parse_dates=['date'])\n",
        "test = pd.read_csv('test.csv', parse_dates=['date'])\n",
        "df = pd.concat([train, test], sort=False)\n",
        "df.head(15)"
      ],
      "id": "090b71f6"
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "089e42af",
        "outputId": "e5dbf5f8-c134-410c-d5e7-6c6decb64a15"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Size of train set (719464, 4)\n",
            "Size of test set: (45000, 4)\n"
          ]
        }
      ],
      "source": [
        "print(\"Size of train set\",train.shape)\n",
        "print(\"Size of test set:\",test.shape)"
      ],
      "id": "089e42af"
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "c79dddae",
        "outputId": "db970bc1-fb92-4f29-818d-0e947323ee74"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Index(['date', 'store', 'item', 'sales'], dtype='object')"
            ]
          },
          "metadata": {},
          "execution_count": 4
        }
      ],
      "source": [
        "#Removing the extra column 'id'\n",
        "df.drop(['id'],inplace=True,axis=1)\n",
        "df.columns"
      ],
      "id": "c79dddae"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "f12a9ee7"
      },
      "source": [
        "## 4. Basic Exploratory data Analysis"
      ],
      "id": "f12a9ee7"
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "84a1aaf8",
        "outputId": "7103385f-942c-4a4b-fee6-7f4373e75495"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Date range: 2013-01-01 00:00:00 to 2018-03-31 00:00:00\n"
          ]
        }
      ],
      "source": [
        "#DATE RANGE\n",
        "\n",
        "print(\"Date range:\", df[\"date\"].min(), \"to\", df[\"date\"].max())\n",
        "#1st Jan 2013 to 31st March, 2018"
      ],
      "id": "84a1aaf8"
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "10690550",
        "outputId": "25a87883-3394-470e-f2bd-116c70ba031c"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "count    719463.000000\n",
              "mean         54.539434\n",
              "std          29.305152\n",
              "min           0.000000\n",
              "10%          21.000000\n",
              "30%          35.000000\n",
              "50%          50.000000\n",
              "70%          67.000000\n",
              "80%          79.000000\n",
              "90%          96.000000\n",
              "95%         110.000000\n",
              "99%         137.000000\n",
              "max         231.000000\n",
              "Name: sales, dtype: float64"
            ]
          },
          "metadata": {},
          "execution_count": 6
        }
      ],
      "source": [
        "# SALES DISTRIBUTION\n",
        "\n",
        "df[\"sales\"].describe([0.10, 0.30, 0.50, 0.70, 0.80, 0.90, 0.95, 0.99])"
      ],
      "id": "10690550"
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "e573958d",
        "outputId": "7cf8ec30-019f-445a-9409-4c2bd5d20883"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "10"
            ]
          },
          "metadata": {},
          "execution_count": 7
        }
      ],
      "source": [
        "# NUMBER OF STORES\n",
        "\n",
        "df[\"store\"].nunique()"
      ],
      "id": "e573958d"
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "3523e380",
        "outputId": "e1ae3838-833a-4064-953f-2e0ab6808782"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "50"
            ]
          },
          "metadata": {},
          "execution_count": 8
        }
      ],
      "source": [
        "# NUMBER OF PRODUCTS\n",
        "\n",
        "df[\"item\"].nunique() "
      ],
      "id": "3523e380"
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "a6dbdbb4",
        "outputId": "9c788e84-e8e8-4659-a81d-9f1311a05fe2"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "store\n",
              "1.0     50\n",
              "2.0     50\n",
              "3.0     50\n",
              "4.0     50\n",
              "5.0     50\n",
              "6.0     50\n",
              "7.0     50\n",
              "8.0     50\n",
              "9.0     50\n",
              "10.0    50\n",
              "Name: item, dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 9
        }
      ],
      "source": [
        "# NUMBER OF PRODUCTS IN EACH STORE\n",
        "df.groupby([\"store\"])[\"item\"].nunique()\n",
        "#Every store sells all the 50 products"
      ],
      "id": "a6dbdbb4"
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 486
        },
        "id": "cd3d99d0",
        "outputId": "e925de63-83ef-407f-a30e-8accb402c066"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "              sales                             \n",
              "                sum       mean median        std\n",
              "store item                                      \n",
              "1.0   1.0   36468.0  19.971522   19.0   6.741022\n",
              "      2.0   97050.0  53.148959   52.0  15.005779\n",
              "      3.0   60638.0  33.208105   33.0  10.072529\n",
              "      4.0   36440.0  19.956188   20.0   6.640618\n",
              "      5.0   30335.0  16.612815   16.0   5.672102\n",
              "...             ...        ...    ...        ...\n",
              "10.0  46.0      0.0        NaN    NaN        NaN\n",
              "      47.0      0.0        NaN    NaN        NaN\n",
              "      48.0      0.0        NaN    NaN        NaN\n",
              "      49.0      0.0        NaN    NaN        NaN\n",
              "      50.0      0.0        NaN    NaN        NaN\n",
              "\n",
              "[500 rows x 4 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-c1f134ca-0f5c-4ac3-9184-d2391c769bde\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead tr th {\n",
              "        text-align: left;\n",
              "    }\n",
              "\n",
              "    .dataframe thead tr:last-of-type th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th colspan=\"4\" halign=\"left\">sales</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th>sum</th>\n",
              "      <th>mean</th>\n",
              "      <th>median</th>\n",
              "      <th>std</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>store</th>\n",
              "      <th>item</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th rowspan=\"5\" valign=\"top\">1.0</th>\n",
              "      <th>1.0</th>\n",
              "      <td>36468.0</td>\n",
              "      <td>19.971522</td>\n",
              "      <td>19.0</td>\n",
              "      <td>6.741022</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2.0</th>\n",
              "      <td>97050.0</td>\n",
              "      <td>53.148959</td>\n",
              "      <td>52.0</td>\n",
              "      <td>15.005779</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3.0</th>\n",
              "      <td>60638.0</td>\n",
              "      <td>33.208105</td>\n",
              "      <td>33.0</td>\n",
              "      <td>10.072529</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4.0</th>\n",
              "      <td>36440.0</td>\n",
              "      <td>19.956188</td>\n",
              "      <td>20.0</td>\n",
              "      <td>6.640618</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5.0</th>\n",
              "      <td>30335.0</td>\n",
              "      <td>16.612815</td>\n",
              "      <td>16.0</td>\n",
              "      <td>5.672102</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th rowspan=\"5\" valign=\"top\">10.0</th>\n",
              "      <th>46.0</th>\n",
              "      <td>0.0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>47.0</th>\n",
              "      <td>0.0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>48.0</th>\n",
              "      <td>0.0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>49.0</th>\n",
              "      <td>0.0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>50.0</th>\n",
              "      <td>0.0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>500 rows × 4 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-c1f134ca-0f5c-4ac3-9184-d2391c769bde')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-c1f134ca-0f5c-4ac3-9184-d2391c769bde button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-c1f134ca-0f5c-4ac3-9184-d2391c769bde');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 10
        }
      ],
      "source": [
        "# Sales statistics in store-product breakdown\n",
        "df.groupby([\"store\", \"item\"]).agg({\"sales\": [\"sum\", \"mean\", \"median\", \"std\"]})"
      ],
      "id": "cd3d99d0"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "f745a3b9"
      },
      "source": [
        "## 5. Feature Engineering"
      ],
      "id": "f745a3b9"
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "metadata": {
        "id": "0ff8c7b1"
      },
      "outputs": [],
      "source": [
        "# Generating date and time parameters from given date\n",
        "\n",
        "df['month'] = df.date.dt.month\n",
        "df['day_of_month'] = df.date.dt.day\n",
        "df['day_of_year'] = df.date.dt.dayofyear \n",
        "df['week_of_year'] = df.date.dt.weekofyear\n",
        "df['day_of_week'] = df.date.dt.dayofweek\n",
        "df['year'] = df.date.dt.year\n",
        "df[\"is_wknd\"] = df.date.dt.weekday // 4\n",
        "df['is_month_start'] = df.date.dt.is_month_start.astype(int)\n",
        "df['is_month_end'] = df.date.dt.is_month_end.astype(int) "
      ],
      "id": "0ff8c7b1"
    },
    {
      "cell_type": "code",
      "execution_count": 12,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 357
        },
        "id": "fad80590",
        "outputId": "348c93bc-0538-4c73-d1b0-7ee2f48889d1"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "        date  store  item  sales  month  day_of_month  day_of_year  week_of_year  day_of_week  year  is_wknd  is_month_start  is_month_end\n",
              "0 2013-01-01    1.0   1.0   13.0      1             1            1             1            1  2013        0               1             0\n",
              "1 2013-01-02    1.0   1.0   11.0      1             2            2             1            2  2013        0               0             0\n",
              "2 2013-01-03    1.0   1.0   14.0      1             3            3             1            3  2013        0               0             0\n",
              "3 2013-01-04    1.0   1.0   13.0      1             4            4             1            4  2013        1               0             0\n",
              "4 2013-01-05    1.0   1.0   10.0      1             5            5             1            5  2013        1               0             0"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-9a9ec3c7-6c31-4691-b951-feb8a4298915\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>date</th>\n",
              "      <th>store</th>\n",
              "      <th>item</th>\n",
              "      <th>sales</th>\n",
              "      <th>month</th>\n",
              "      <th>day_of_month</th>\n",
              "      <th>day_of_year</th>\n",
              "      <th>week_of_year</th>\n",
              "      <th>day_of_week</th>\n",
              "      <th>year</th>\n",
              "      <th>is_wknd</th>\n",
              "      <th>is_month_start</th>\n",
              "      <th>is_month_end</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>2013-01-01</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>13.0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>2013</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>2013-01-02</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>11.0</td>\n",
              "      <td>1</td>\n",
              "      <td>2</td>\n",
              "      <td>2</td>\n",
              "      <td>1</td>\n",
              "      <td>2</td>\n",
              "      <td>2013</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2013-01-03</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>14.0</td>\n",
              "      <td>1</td>\n",
              "      <td>3</td>\n",
              "      <td>3</td>\n",
              "      <td>1</td>\n",
              "      <td>3</td>\n",
              "      <td>2013</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>2013-01-04</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>13.0</td>\n",
              "      <td>1</td>\n",
              "      <td>4</td>\n",
              "      <td>4</td>\n",
              "      <td>1</td>\n",
              "      <td>4</td>\n",
              "      <td>2013</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>2013-01-05</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>10.0</td>\n",
              "      <td>1</td>\n",
              "      <td>5</td>\n",
              "      <td>5</td>\n",
              "      <td>1</td>\n",
              "      <td>5</td>\n",
              "      <td>2013</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-9a9ec3c7-6c31-4691-b951-feb8a4298915')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-9a9ec3c7-6c31-4691-b951-feb8a4298915 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-9a9ec3c7-6c31-4691-b951-feb8a4298915');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 12
        }
      ],
      "source": [
        "df.head()"
      ],
      "id": "fad80590"
    },
    {
      "cell_type": "code",
      "execution_count": 13,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 486
        },
        "id": "089de2f6",
        "outputId": "55ec3f7a-9694-4cd3-8f4e-6302833c939b"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "                   sales                            \n",
              "                     sum       mean median       std\n",
              "store item month                                    \n",
              "1.0   1.0  1      2125.0  13.709677   13.0  4.397413\n",
              "           2      2063.0  14.631206   14.0  4.668146\n",
              "           3      2728.0  17.600000   17.0  4.545013\n",
              "           4      3118.0  20.786667   20.0  4.894301\n",
              "           5      3448.0  22.245161   22.0  6.564705\n",
              "...                  ...        ...    ...       ...\n",
              "10.0  49.0 2         0.0        NaN    NaN       NaN\n",
              "           3         0.0        NaN    NaN       NaN\n",
              "      50.0 1         0.0        NaN    NaN       NaN\n",
              "           2         0.0        NaN    NaN       NaN\n",
              "           3         0.0        NaN    NaN       NaN\n",
              "\n",
              "[5046 rows x 4 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-840ba8bf-d1d4-4dfb-b4bb-f4e7772186fb\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead tr th {\n",
              "        text-align: left;\n",
              "    }\n",
              "\n",
              "    .dataframe thead tr:last-of-type th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th colspan=\"4\" halign=\"left\">sales</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th>sum</th>\n",
              "      <th>mean</th>\n",
              "      <th>median</th>\n",
              "      <th>std</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>store</th>\n",
              "      <th>item</th>\n",
              "      <th>month</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th rowspan=\"5\" valign=\"top\">1.0</th>\n",
              "      <th rowspan=\"5\" valign=\"top\">1.0</th>\n",
              "      <th>1</th>\n",
              "      <td>2125.0</td>\n",
              "      <td>13.709677</td>\n",
              "      <td>13.0</td>\n",
              "      <td>4.397413</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2063.0</td>\n",
              "      <td>14.631206</td>\n",
              "      <td>14.0</td>\n",
              "      <td>4.668146</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>2728.0</td>\n",
              "      <td>17.600000</td>\n",
              "      <td>17.0</td>\n",
              "      <td>4.545013</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>3118.0</td>\n",
              "      <td>20.786667</td>\n",
              "      <td>20.0</td>\n",
              "      <td>4.894301</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>3448.0</td>\n",
              "      <td>22.245161</td>\n",
              "      <td>22.0</td>\n",
              "      <td>6.564705</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <th>...</th>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th rowspan=\"5\" valign=\"top\">10.0</th>\n",
              "      <th rowspan=\"2\" valign=\"top\">49.0</th>\n",
              "      <th>2</th>\n",
              "      <td>0.0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0.0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th rowspan=\"3\" valign=\"top\">50.0</th>\n",
              "      <th>1</th>\n",
              "      <td>0.0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0.0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0.0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>5046 rows × 4 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-840ba8bf-d1d4-4dfb-b4bb-f4e7772186fb')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-840ba8bf-d1d4-4dfb-b4bb-f4e7772186fb button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-840ba8bf-d1d4-4dfb-b4bb-f4e7772186fb');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 13
        }
      ],
      "source": [
        "# Sales statistics in store-item-month breakdown\n",
        "df.groupby([\"store\", \"item\", \"month\"]).agg({\"sales\": [\"sum\", \"mean\", \"median\", \"std\"]})"
      ],
      "id": "089de2f6"
    },
    {
      "cell_type": "code",
      "execution_count": 14,
      "metadata": {
        "id": "bf37f111"
      },
      "outputs": [],
      "source": [
        "#DEALING WITH RAMDOM NOISE\n",
        "# For small datasets like this dataset, random noise can be added to the values ​​to prevent overfitting.\n",
        "# Here I will add Gaussian random noise, which is normally distributed with 1 standard deviation and 0 mean.\n",
        "\n",
        "def random_noise(dataframe):\n",
        "    return np.random.normal(scale=1.6, size=(len(dataframe),))"
      ],
      "id": "bf37f111"
    },
    {
      "cell_type": "code",
      "execution_count": 15,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 601
        },
        "id": "9b3ae8d6",
        "outputId": "c4e8fb99-1f01-4687-c70f-1e49412381eb"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "        date  store  item  sales  month  day_of_month  day_of_year  week_of_year  day_of_week  year  is_wknd  is_month_start  is_month_end\n",
              "0 2013-01-01    1.0   1.0   13.0      1             1            1             1            1  2013        0               1             0\n",
              "1 2013-01-02    1.0   1.0   11.0      1             2            2             1            2  2013        0               0             0\n",
              "2 2013-01-03    1.0   1.0   14.0      1             3            3             1            3  2013        0               0             0\n",
              "3 2013-01-04    1.0   1.0   13.0      1             4            4             1            4  2013        1               0             0\n",
              "4 2013-01-05    1.0   1.0   10.0      1             5            5             1            5  2013        1               0             0\n",
              "5 2013-01-06    1.0   1.0   12.0      1             6            6             1            6  2013        1               0             0\n",
              "6 2013-01-07    1.0   1.0   10.0      1             7            7             2            0  2013        0               0             0\n",
              "7 2013-01-08    1.0   1.0    9.0      1             8            8             2            1  2013        0               0             0\n",
              "8 2013-01-09    1.0   1.0   12.0      1             9            9             2            2  2013        0               0             0\n",
              "9 2013-01-10    1.0   1.0    9.0      1            10           10             2            3  2013        0               0             0"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-33fdbd8f-e51a-45c7-8074-aed69453d360\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>date</th>\n",
              "      <th>store</th>\n",
              "      <th>item</th>\n",
              "      <th>sales</th>\n",
              "      <th>month</th>\n",
              "      <th>day_of_month</th>\n",
              "      <th>day_of_year</th>\n",
              "      <th>week_of_year</th>\n",
              "      <th>day_of_week</th>\n",
              "      <th>year</th>\n",
              "      <th>is_wknd</th>\n",
              "      <th>is_month_start</th>\n",
              "      <th>is_month_end</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>2013-01-01</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>13.0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>2013</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>2013-01-02</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>11.0</td>\n",
              "      <td>1</td>\n",
              "      <td>2</td>\n",
              "      <td>2</td>\n",
              "      <td>1</td>\n",
              "      <td>2</td>\n",
              "      <td>2013</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2013-01-03</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>14.0</td>\n",
              "      <td>1</td>\n",
              "      <td>3</td>\n",
              "      <td>3</td>\n",
              "      <td>1</td>\n",
              "      <td>3</td>\n",
              "      <td>2013</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>2013-01-04</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>13.0</td>\n",
              "      <td>1</td>\n",
              "      <td>4</td>\n",
              "      <td>4</td>\n",
              "      <td>1</td>\n",
              "      <td>4</td>\n",
              "      <td>2013</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>2013-01-05</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>10.0</td>\n",
              "      <td>1</td>\n",
              "      <td>5</td>\n",
              "      <td>5</td>\n",
              "      <td>1</td>\n",
              "      <td>5</td>\n",
              "      <td>2013</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>2013-01-06</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>12.0</td>\n",
              "      <td>1</td>\n",
              "      <td>6</td>\n",
              "      <td>6</td>\n",
              "      <td>1</td>\n",
              "      <td>6</td>\n",
              "      <td>2013</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>2013-01-07</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>10.0</td>\n",
              "      <td>1</td>\n",
              "      <td>7</td>\n",
              "      <td>7</td>\n",
              "      <td>2</td>\n",
              "      <td>0</td>\n",
              "      <td>2013</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>2013-01-08</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>9.0</td>\n",
              "      <td>1</td>\n",
              "      <td>8</td>\n",
              "      <td>8</td>\n",
              "      <td>2</td>\n",
              "      <td>1</td>\n",
              "      <td>2013</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>2013-01-09</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>12.0</td>\n",
              "      <td>1</td>\n",
              "      <td>9</td>\n",
              "      <td>9</td>\n",
              "      <td>2</td>\n",
              "      <td>2</td>\n",
              "      <td>2013</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>2013-01-10</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>9.0</td>\n",
              "      <td>1</td>\n",
              "      <td>10</td>\n",
              "      <td>10</td>\n",
              "      <td>2</td>\n",
              "      <td>3</td>\n",
              "      <td>2013</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-33fdbd8f-e51a-45c7-8074-aed69453d360')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-33fdbd8f-e51a-45c7-8074-aed69453d360 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-33fdbd8f-e51a-45c7-8074-aed69453d360');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 15
        }
      ],
      "source": [
        "#Lag/Shifted Features (Delays)\n",
        "df.sort_values(by=['store', 'item', 'date'], axis=0, inplace=True)\n",
        "df.head(10)"
      ],
      "id": "9b3ae8d6"
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "metadata": {
        "id": "2524ec4f"
      },
      "outputs": [],
      "source": [
        "def lag_features(dataframe, lags):\n",
        "    for lag in lags:\n",
        "        dataframe['sales_lag_' + str(lag)] = dataframe.groupby([\"store\", \"item\"])['sales'].transform(\n",
        "            lambda x: x.shift(lag)) + random_noise(dataframe)\n",
        "    return dataframe\n",
        "\n",
        "df = lag_features(df, [91, 98, 105, 112, 119, 126, 182, 364, 546, 728])"
      ],
      "id": "2524ec4f"
    },
    {
      "cell_type": "code",
      "execution_count": 17,
      "metadata": {
        "id": "3a43ca79"
      },
      "outputs": [],
      "source": [
        "#Moving Average Features\n",
        "def moving_average_features(dataframe, windows):\n",
        "    for window in windows:\n",
        "        dataframe['sales_roll_mean_' + str(window)] = dataframe.groupby([\"store\", \"item\"])['sales']. \\\n",
        "                                                          transform(\n",
        "            lambda x: x.shift(1).rolling(window=window, min_periods=10, win_type=\"triang\").mean()) + random_noise(\n",
        "            dataframe)\n",
        "    return dataframe\n",
        "\n",
        "\n",
        "df = moving_average_features(df, [365, 546, 730])\n"
      ],
      "id": "3a43ca79"
    },
    {
      "cell_type": "code",
      "execution_count": 18,
      "metadata": {
        "id": "959e7059"
      },
      "outputs": [],
      "source": [
        "#Exponentially Weighted Average Features\n",
        "def ewm_features(dataframe, alphas, lags):\n",
        "    for alpha in alphas:\n",
        "        for lag in lags:\n",
        "            dataframe['sales_ewm_alpha_' + str(alpha).replace(\".\", \"\") + \"_lag_\" + str(lag)] = \\\n",
        "                dataframe.groupby([\"store\", \"item\"])['sales'].transform(lambda x: x.shift(lag).ewm(alpha=alpha).mean())\n",
        "    return dataframe\n",
        "\n",
        "\n",
        "alphas = [0.99, 0.95, 0.9, 0.8, 0.7, 0.5]\n",
        "lags = [91, 98, 105, 112, 180, 270, 365, 546, 728]\n",
        "\n",
        "df = ewm_features(df, alphas, lags)"
      ],
      "id": "959e7059"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "eee97bd4"
      },
      "source": [
        "## 6. Data Encoding"
      ],
      "id": "eee97bd4"
    },
    {
      "cell_type": "code",
      "execution_count": 19,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4c3a759f",
        "outputId": "04ce07d4-ff53-4309-b3f3-3ce67fa6910a"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "Int64Index: 764464 entries, 0 to 719463\n",
            "Data columns (total 80 columns):\n",
            " #   Column                       Non-Null Count   Dtype         \n",
            "---  ------                       --------------   -----         \n",
            " 0   date                         764464 non-null  datetime64[ns]\n",
            " 1   store                        764463 non-null  float64       \n",
            " 2   item                         764463 non-null  float64       \n",
            " 3   sales                        719463 non-null  float64       \n",
            " 4   month                        764464 non-null  int64         \n",
            " 5   day_of_month                 764464 non-null  int64         \n",
            " 6   day_of_year                  764464 non-null  int64         \n",
            " 7   week_of_year                 764464 non-null  int64         \n",
            " 8   day_of_week                  764464 non-null  int64         \n",
            " 9   year                         764464 non-null  int64         \n",
            " 10  is_wknd                      764464 non-null  int64         \n",
            " 11  is_month_start               764464 non-null  int64         \n",
            " 12  is_month_end                 764464 non-null  int64         \n",
            " 13  sales_lag_91                 719068 non-null  float64       \n",
            " 14  sales_lag_98                 716303 non-null  float64       \n",
            " 15  sales_lag_105                713538 non-null  float64       \n",
            " 16  sales_lag_112                710776 non-null  float64       \n",
            " 17  sales_lag_119                708018 non-null  float64       \n",
            " 18  sales_lag_126                705260 non-null  float64       \n",
            " 19  sales_lag_182                683196 non-null  float64       \n",
            " 20  sales_lag_364                611488 non-null  float64       \n",
            " 21  sales_lag_546                539780 non-null  float64       \n",
            " 22  sales_lag_728                468072 non-null  float64       \n",
            " 23  sales_roll_mean_365          751063 non-null  float64       \n",
            " 24  sales_roll_mean_546          751063 non-null  float64       \n",
            " 25  sales_roll_mean_730          751063 non-null  float64       \n",
            " 26  sales_ewm_alpha_099_lag_91   719068 non-null  float64       \n",
            " 27  sales_ewm_alpha_099_lag_98   716303 non-null  float64       \n",
            " 28  sales_ewm_alpha_099_lag_105  713538 non-null  float64       \n",
            " 29  sales_ewm_alpha_099_lag_112  710776 non-null  float64       \n",
            " 30  sales_ewm_alpha_099_lag_180  683984 non-null  float64       \n",
            " 31  sales_ewm_alpha_099_lag_270  648524 non-null  float64       \n",
            " 32  sales_ewm_alpha_099_lag_365  611094 non-null  float64       \n",
            " 33  sales_ewm_alpha_099_lag_546  539780 non-null  float64       \n",
            " 34  sales_ewm_alpha_099_lag_728  468072 non-null  float64       \n",
            " 35  sales_ewm_alpha_095_lag_91   719068 non-null  float64       \n",
            " 36  sales_ewm_alpha_095_lag_98   716303 non-null  float64       \n",
            " 37  sales_ewm_alpha_095_lag_105  713538 non-null  float64       \n",
            " 38  sales_ewm_alpha_095_lag_112  710776 non-null  float64       \n",
            " 39  sales_ewm_alpha_095_lag_180  683984 non-null  float64       \n",
            " 40  sales_ewm_alpha_095_lag_270  648524 non-null  float64       \n",
            " 41  sales_ewm_alpha_095_lag_365  611094 non-null  float64       \n",
            " 42  sales_ewm_alpha_095_lag_546  539780 non-null  float64       \n",
            " 43  sales_ewm_alpha_095_lag_728  468072 non-null  float64       \n",
            " 44  sales_ewm_alpha_09_lag_91    719068 non-null  float64       \n",
            " 45  sales_ewm_alpha_09_lag_98    716303 non-null  float64       \n",
            " 46  sales_ewm_alpha_09_lag_105   713538 non-null  float64       \n",
            " 47  sales_ewm_alpha_09_lag_112   710776 non-null  float64       \n",
            " 48  sales_ewm_alpha_09_lag_180   683984 non-null  float64       \n",
            " 49  sales_ewm_alpha_09_lag_270   648524 non-null  float64       \n",
            " 50  sales_ewm_alpha_09_lag_365   611094 non-null  float64       \n",
            " 51  sales_ewm_alpha_09_lag_546   539780 non-null  float64       \n",
            " 52  sales_ewm_alpha_09_lag_728   468072 non-null  float64       \n",
            " 53  sales_ewm_alpha_08_lag_91    719068 non-null  float64       \n",
            " 54  sales_ewm_alpha_08_lag_98    716303 non-null  float64       \n",
            " 55  sales_ewm_alpha_08_lag_105   713538 non-null  float64       \n",
            " 56  sales_ewm_alpha_08_lag_112   710776 non-null  float64       \n",
            " 57  sales_ewm_alpha_08_lag_180   683984 non-null  float64       \n",
            " 58  sales_ewm_alpha_08_lag_270   648524 non-null  float64       \n",
            " 59  sales_ewm_alpha_08_lag_365   611094 non-null  float64       \n",
            " 60  sales_ewm_alpha_08_lag_546   539780 non-null  float64       \n",
            " 61  sales_ewm_alpha_08_lag_728   468072 non-null  float64       \n",
            " 62  sales_ewm_alpha_07_lag_91    719068 non-null  float64       \n",
            " 63  sales_ewm_alpha_07_lag_98    716303 non-null  float64       \n",
            " 64  sales_ewm_alpha_07_lag_105   713538 non-null  float64       \n",
            " 65  sales_ewm_alpha_07_lag_112   710776 non-null  float64       \n",
            " 66  sales_ewm_alpha_07_lag_180   683984 non-null  float64       \n",
            " 67  sales_ewm_alpha_07_lag_270   648524 non-null  float64       \n",
            " 68  sales_ewm_alpha_07_lag_365   611094 non-null  float64       \n",
            " 69  sales_ewm_alpha_07_lag_546   539780 non-null  float64       \n",
            " 70  sales_ewm_alpha_07_lag_728   468072 non-null  float64       \n",
            " 71  sales_ewm_alpha_05_lag_91    719068 non-null  float64       \n",
            " 72  sales_ewm_alpha_05_lag_98    716303 non-null  float64       \n",
            " 73  sales_ewm_alpha_05_lag_105   713538 non-null  float64       \n",
            " 74  sales_ewm_alpha_05_lag_112   710776 non-null  float64       \n",
            " 75  sales_ewm_alpha_05_lag_180   683984 non-null  float64       \n",
            " 76  sales_ewm_alpha_05_lag_270   648524 non-null  float64       \n",
            " 77  sales_ewm_alpha_05_lag_365   611094 non-null  float64       \n",
            " 78  sales_ewm_alpha_05_lag_546   539780 non-null  float64       \n",
            " 79  sales_ewm_alpha_05_lag_728   468072 non-null  float64       \n",
            "dtypes: datetime64[ns](1), float64(70), int64(9)\n",
            "memory usage: 472.4 MB\n"
          ]
        }
      ],
      "source": [
        "#Checking for null values\n",
        "df.info()"
      ],
      "id": "4c3a759f"
    },
    {
      "cell_type": "code",
      "execution_count": 20,
      "metadata": {
        "id": "2a8b0531"
      },
      "outputs": [],
      "source": [
        "#One-Hot Encoding\n",
        "df = pd.get_dummies(df, columns=['day_of_week', 'month'])"
      ],
      "id": "2a8b0531"
    },
    {
      "cell_type": "code",
      "execution_count": 21,
      "metadata": {
        "id": "d4e89f99"
      },
      "outputs": [],
      "source": [
        "#Converting sales to log(1+sales)\n",
        "df['sales'] = np.log1p(df[\"sales\"].values)"
      ],
      "id": "d4e89f99"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "a6995824"
      },
      "source": [
        "## 7. LightGBM Model"
      ],
      "id": "a6995824"
    },
    {
      "cell_type": "code",
      "execution_count": 22,
      "metadata": {
        "id": "b90334ff"
      },
      "outputs": [],
      "source": [
        "# Train data set until the beginning of 2017 (end of 2016)\n",
        "train = df.loc[(df[\"date\"] < \"2017-01-01\"), :]\n",
        "\n",
        "# First 3 months of 2017 validation set\n",
        "val = df.loc[(df[\"date\"] >= \"2017-01-01\") & (df[\"date\"] < \"2017-04-01\"), :]\n",
        "\n",
        "# Independent variables\n",
        "cols = [col for col in train.columns if col not in ['date', 'id', \"sales\", \"year\"]]"
      ],
      "id": "b90334ff"
    },
    {
      "cell_type": "code",
      "execution_count": 23,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7aad7fe7",
        "outputId": "2011d808-a208-4fcb-cb12-6777a140e8cc"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((575654,), (575654, 94), (35460,), (35460, 94))"
            ]
          },
          "metadata": {},
          "execution_count": 23
        }
      ],
      "source": [
        "# Selecting the dependent variable for the train set\n",
        "Y_train = train['sales']\n",
        "\n",
        "# Selecting the argument for the train set\n",
        "X_train = train[cols]\n",
        "\n",
        "# choosing the dependent variable for the validation set\n",
        "Y_val = val['sales']\n",
        "\n",
        "# selecting the independent variable for the validation set\n",
        "X_val = val[cols] \n",
        "\n",
        "# Checking the shapes\n",
        "Y_train.shape, X_train.shape, Y_val.shape, X_val.shape"
      ],
      "id": "7aad7fe7"
    },
    {
      "cell_type": "code",
      "execution_count": 24,
      "metadata": {
        "id": "a5d2c181"
      },
      "outputs": [],
      "source": [
        "# Custom Cost Function\n",
        "\n",
        "def smape(preds, target):\n",
        "    n = len(preds)\n",
        "    masked_arr = ~((preds == 0) & (target == 0))\n",
        "    preds, target = preds[masked_arr], target[masked_arr]\n",
        "    num = np.abs(preds - target)\n",
        "    denom = np.abs(preds) + np.abs(target)\n",
        "    smape_val = (200 * np.sum(num / denom)) / n\n",
        "    return smape_val\n",
        "\n",
        "\n",
        "def lgbm_smape(preds, train_data):\n",
        "    labels = train_data.get_label()\n",
        "    smape_val = smape(np.expm1(preds), np.expm1(labels))\n",
        "    return 'SMAPE', smape_val, False\n"
      ],
      "id": "a5d2c181"
    },
    {
      "cell_type": "code",
      "execution_count": 25,
      "metadata": {
        "id": "3040e35f"
      },
      "outputs": [],
      "source": [
        "# LightGBM parameters\n",
        "lgb_params = {'metric': {'mae'},\n",
        "              'num_leaves': 10,\n",
        "              'learning_rate': 0.02,\n",
        "              'feature_fraction': 0.8,\n",
        "              'max_depth': 5,\n",
        "              'verbose': 0,\n",
        "              'num_boost_round': 10000, \n",
        "              'early_stopping_rounds': 200,\n",
        "              'nthread': -1}"
      ],
      "id": "3040e35f"
    },
    {
      "cell_type": "code",
      "execution_count": 26,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "f7890a9d",
        "outputId": "843a4904-c2d0-4569-e5ce-b17115d67df5"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[LightGBM] [Warning] Auto-choosing col-wise multi-threading, the overhead of testing was 0.350047 seconds.\n",
            "You can set `force_col_wise=true` to remove the overhead.\n",
            "Training until validation scores don't improve for 200 rounds\n",
            "[100]\ttraining's l1: 0.168608\ttraining's SMAPE: 17.179\tvalid_1's l1: 0.167998\tvalid_1's SMAPE: 17.1546\n",
            "[200]\ttraining's l1: 0.138493\ttraining's SMAPE: 14.1696\tvalid_1's l1: 0.141842\tvalid_1's SMAPE: 14.5346\n",
            "[300]\ttraining's l1: 0.13292\ttraining's SMAPE: 13.6118\tvalid_1's l1: 0.136922\tvalid_1's SMAPE: 14.0404\n",
            "[400]\ttraining's l1: 0.130788\ttraining's SMAPE: 13.3988\tvalid_1's l1: 0.135205\tvalid_1's SMAPE: 13.8681\n",
            "[500]\ttraining's l1: 0.129528\ttraining's SMAPE: 13.2726\tvalid_1's l1: 0.133886\tvalid_1's SMAPE: 13.7355\n",
            "[600]\ttraining's l1: 0.128685\ttraining's SMAPE: 13.1879\tvalid_1's l1: 0.132911\tvalid_1's SMAPE: 13.6374\n",
            "[700]\ttraining's l1: 0.128069\ttraining's SMAPE: 13.126\tvalid_1's l1: 0.132125\tvalid_1's SMAPE: 13.5582\n",
            "[800]\ttraining's l1: 0.127583\ttraining's SMAPE: 13.0772\tvalid_1's l1: 0.13154\tvalid_1's SMAPE: 13.4994\n",
            "[900]\ttraining's l1: 0.127161\ttraining's SMAPE: 13.035\tvalid_1's l1: 0.131099\tvalid_1's SMAPE: 13.455\n",
            "[1000]\ttraining's l1: 0.126799\ttraining's SMAPE: 12.9988\tvalid_1's l1: 0.130755\tvalid_1's SMAPE: 13.4203\n",
            "[1100]\ttraining's l1: 0.126505\ttraining's SMAPE: 12.9694\tvalid_1's l1: 0.130503\tvalid_1's SMAPE: 13.3949\n",
            "[1200]\ttraining's l1: 0.126263\ttraining's SMAPE: 12.9451\tvalid_1's l1: 0.130278\tvalid_1's SMAPE: 13.3722\n",
            "[1300]\ttraining's l1: 0.126023\ttraining's SMAPE: 12.9211\tvalid_1's l1: 0.130074\tvalid_1's SMAPE: 13.3516\n",
            "[1400]\ttraining's l1: 0.12582\ttraining's SMAPE: 12.9008\tvalid_1's l1: 0.129904\tvalid_1's SMAPE: 13.3345\n",
            "[1500]\ttraining's l1: 0.125645\ttraining's SMAPE: 12.8832\tvalid_1's l1: 0.129769\tvalid_1's SMAPE: 13.3209\n",
            "[1600]\ttraining's l1: 0.125493\ttraining's SMAPE: 12.868\tvalid_1's l1: 0.129646\tvalid_1's SMAPE: 13.3085\n",
            "[1700]\ttraining's l1: 0.125354\ttraining's SMAPE: 12.854\tvalid_1's l1: 0.129536\tvalid_1's SMAPE: 13.2975\n",
            "[1800]\ttraining's l1: 0.125233\ttraining's SMAPE: 12.8418\tvalid_1's l1: 0.129435\tvalid_1's SMAPE: 13.2873\n",
            "[1900]\ttraining's l1: 0.125113\ttraining's SMAPE: 12.8297\tvalid_1's l1: 0.129356\tvalid_1's SMAPE: 13.2794\n",
            "[2000]\ttraining's l1: 0.124989\ttraining's SMAPE: 12.8173\tvalid_1's l1: 0.129284\tvalid_1's SMAPE: 13.2721\n",
            "[2100]\ttraining's l1: 0.124884\ttraining's SMAPE: 12.8068\tvalid_1's l1: 0.129207\tvalid_1's SMAPE: 13.2643\n",
            "[2200]\ttraining's l1: 0.12478\ttraining's SMAPE: 12.7964\tvalid_1's l1: 0.129149\tvalid_1's SMAPE: 13.2585\n",
            "[2300]\ttraining's l1: 0.124686\ttraining's SMAPE: 12.7869\tvalid_1's l1: 0.129097\tvalid_1's SMAPE: 13.2533\n",
            "[2400]\ttraining's l1: 0.12459\ttraining's SMAPE: 12.7773\tvalid_1's l1: 0.129047\tvalid_1's SMAPE: 13.2483\n",
            "[2500]\ttraining's l1: 0.124492\ttraining's SMAPE: 12.7675\tvalid_1's l1: 0.128988\tvalid_1's SMAPE: 13.2423\n",
            "[2600]\ttraining's l1: 0.124404\ttraining's SMAPE: 12.7587\tvalid_1's l1: 0.128945\tvalid_1's SMAPE: 13.238\n",
            "[2700]\ttraining's l1: 0.12432\ttraining's SMAPE: 12.7503\tvalid_1's l1: 0.128893\tvalid_1's SMAPE: 13.2328\n",
            "[2800]\ttraining's l1: 0.124241\ttraining's SMAPE: 12.7424\tvalid_1's l1: 0.128845\tvalid_1's SMAPE: 13.2279\n",
            "[2900]\ttraining's l1: 0.124168\ttraining's SMAPE: 12.735\tvalid_1's l1: 0.128799\tvalid_1's SMAPE: 13.2233\n",
            "[3000]\ttraining's l1: 0.124097\ttraining's SMAPE: 12.7278\tvalid_1's l1: 0.128779\tvalid_1's SMAPE: 13.2212\n",
            "[3100]\ttraining's l1: 0.12403\ttraining's SMAPE: 12.721\tvalid_1's l1: 0.128749\tvalid_1's SMAPE: 13.2183\n",
            "[3200]\ttraining's l1: 0.123962\ttraining's SMAPE: 12.7141\tvalid_1's l1: 0.128723\tvalid_1's SMAPE: 13.2156\n",
            "[3300]\ttraining's l1: 0.123891\ttraining's SMAPE: 12.707\tvalid_1's l1: 0.128693\tvalid_1's SMAPE: 13.2126\n",
            "[3400]\ttraining's l1: 0.12382\ttraining's SMAPE: 12.6999\tvalid_1's l1: 0.128655\tvalid_1's SMAPE: 13.2089\n",
            "[3500]\ttraining's l1: 0.123754\ttraining's SMAPE: 12.6932\tvalid_1's l1: 0.128628\tvalid_1's SMAPE: 13.2061\n",
            "[3600]\ttraining's l1: 0.12369\ttraining's SMAPE: 12.6868\tvalid_1's l1: 0.128607\tvalid_1's SMAPE: 13.204\n",
            "[3700]\ttraining's l1: 0.12363\ttraining's SMAPE: 12.6807\tvalid_1's l1: 0.12859\tvalid_1's SMAPE: 13.2023\n",
            "[3800]\ttraining's l1: 0.123572\ttraining's SMAPE: 12.6748\tvalid_1's l1: 0.128565\tvalid_1's SMAPE: 13.1999\n",
            "[3900]\ttraining's l1: 0.123514\ttraining's SMAPE: 12.669\tvalid_1's l1: 0.128529\tvalid_1's SMAPE: 13.1962\n",
            "[4000]\ttraining's l1: 0.123448\ttraining's SMAPE: 12.6623\tvalid_1's l1: 0.128508\tvalid_1's SMAPE: 13.1941\n",
            "[4100]\ttraining's l1: 0.123388\ttraining's SMAPE: 12.6563\tvalid_1's l1: 0.128495\tvalid_1's SMAPE: 13.1928\n",
            "[4200]\ttraining's l1: 0.123332\ttraining's SMAPE: 12.6506\tvalid_1's l1: 0.128471\tvalid_1's SMAPE: 13.1904\n",
            "[4300]\ttraining's l1: 0.123273\ttraining's SMAPE: 12.6446\tvalid_1's l1: 0.128456\tvalid_1's SMAPE: 13.1889\n",
            "[4400]\ttraining's l1: 0.123212\ttraining's SMAPE: 12.6385\tvalid_1's l1: 0.128433\tvalid_1's SMAPE: 13.1866\n",
            "[4500]\ttraining's l1: 0.123154\ttraining's SMAPE: 12.6326\tvalid_1's l1: 0.128412\tvalid_1's SMAPE: 13.1844\n",
            "[4600]\ttraining's l1: 0.123101\ttraining's SMAPE: 12.6273\tvalid_1's l1: 0.128404\tvalid_1's SMAPE: 13.1836\n",
            "[4700]\ttraining's l1: 0.123051\ttraining's SMAPE: 12.6222\tvalid_1's l1: 0.128385\tvalid_1's SMAPE: 13.1818\n",
            "[4800]\ttraining's l1: 0.123002\ttraining's SMAPE: 12.6172\tvalid_1's l1: 0.128371\tvalid_1's SMAPE: 13.1803\n",
            "[4900]\ttraining's l1: 0.122952\ttraining's SMAPE: 12.6121\tvalid_1's l1: 0.12835\tvalid_1's SMAPE: 13.1782\n",
            "[5000]\ttraining's l1: 0.122905\ttraining's SMAPE: 12.6074\tvalid_1's l1: 0.128337\tvalid_1's SMAPE: 13.1769\n",
            "[5100]\ttraining's l1: 0.122851\ttraining's SMAPE: 12.6019\tvalid_1's l1: 0.128323\tvalid_1's SMAPE: 13.1755\n",
            "[5200]\ttraining's l1: 0.1228\ttraining's SMAPE: 12.5967\tvalid_1's l1: 0.12831\tvalid_1's SMAPE: 13.1742\n",
            "[5300]\ttraining's l1: 0.122758\ttraining's SMAPE: 12.5924\tvalid_1's l1: 0.128299\tvalid_1's SMAPE: 13.1731\n",
            "[5400]\ttraining's l1: 0.122713\ttraining's SMAPE: 12.5879\tvalid_1's l1: 0.128283\tvalid_1's SMAPE: 13.1715\n",
            "[5500]\ttraining's l1: 0.122667\ttraining's SMAPE: 12.5832\tvalid_1's l1: 0.12827\tvalid_1's SMAPE: 13.1702\n",
            "[5600]\ttraining's l1: 0.122624\ttraining's SMAPE: 12.5788\tvalid_1's l1: 0.128262\tvalid_1's SMAPE: 13.1694\n",
            "[5700]\ttraining's l1: 0.122576\ttraining's SMAPE: 12.5739\tvalid_1's l1: 0.128241\tvalid_1's SMAPE: 13.1673\n",
            "[5800]\ttraining's l1: 0.122536\ttraining's SMAPE: 12.5699\tvalid_1's l1: 0.128225\tvalid_1's SMAPE: 13.1657\n",
            "[5900]\ttraining's l1: 0.122494\ttraining's SMAPE: 12.5656\tvalid_1's l1: 0.128221\tvalid_1's SMAPE: 13.1653\n",
            "[6000]\ttraining's l1: 0.122454\ttraining's SMAPE: 12.5615\tvalid_1's l1: 0.128218\tvalid_1's SMAPE: 13.165\n",
            "[6100]\ttraining's l1: 0.122411\ttraining's SMAPE: 12.5572\tvalid_1's l1: 0.128211\tvalid_1's SMAPE: 13.1643\n",
            "[6200]\ttraining's l1: 0.122372\ttraining's SMAPE: 12.5532\tvalid_1's l1: 0.128198\tvalid_1's SMAPE: 13.163\n",
            "[6300]\ttraining's l1: 0.12233\ttraining's SMAPE: 12.5489\tvalid_1's l1: 0.128194\tvalid_1's SMAPE: 13.1627\n",
            "[6400]\ttraining's l1: 0.122289\ttraining's SMAPE: 12.5447\tvalid_1's l1: 0.128184\tvalid_1's SMAPE: 13.1617\n",
            "[6500]\ttraining's l1: 0.122245\ttraining's SMAPE: 12.5403\tvalid_1's l1: 0.128179\tvalid_1's SMAPE: 13.1611\n",
            "[6600]\ttraining's l1: 0.122207\ttraining's SMAPE: 12.5364\tvalid_1's l1: 0.12817\tvalid_1's SMAPE: 13.1603\n",
            "[6700]\ttraining's l1: 0.122165\ttraining's SMAPE: 12.5322\tvalid_1's l1: 0.128164\tvalid_1's SMAPE: 13.1596\n",
            "[6800]\ttraining's l1: 0.122125\ttraining's SMAPE: 12.528\tvalid_1's l1: 0.128163\tvalid_1's SMAPE: 13.1596\n",
            "[6900]\ttraining's l1: 0.122085\ttraining's SMAPE: 12.524\tvalid_1's l1: 0.128154\tvalid_1's SMAPE: 13.1586\n",
            "[7000]\ttraining's l1: 0.122046\ttraining's SMAPE: 12.52\tvalid_1's l1: 0.128146\tvalid_1's SMAPE: 13.1579\n",
            "[7100]\ttraining's l1: 0.121997\ttraining's SMAPE: 12.5151\tvalid_1's l1: 0.128134\tvalid_1's SMAPE: 13.1566\n",
            "[7200]\ttraining's l1: 0.121957\ttraining's SMAPE: 12.511\tvalid_1's l1: 0.128133\tvalid_1's SMAPE: 13.1565\n",
            "[7300]\ttraining's l1: 0.121912\ttraining's SMAPE: 12.5064\tvalid_1's l1: 0.128124\tvalid_1's SMAPE: 13.1556\n",
            "[7400]\ttraining's l1: 0.121873\ttraining's SMAPE: 12.5025\tvalid_1's l1: 0.128112\tvalid_1's SMAPE: 13.1544\n",
            "[7500]\ttraining's l1: 0.121836\ttraining's SMAPE: 12.4987\tvalid_1's l1: 0.128116\tvalid_1's SMAPE: 13.1548\n",
            "[7600]\ttraining's l1: 0.121796\ttraining's SMAPE: 12.4946\tvalid_1's l1: 0.128118\tvalid_1's SMAPE: 13.155\n",
            "Early stopping, best iteration is:\n",
            "[7404]\ttraining's l1: 0.121872\ttraining's SMAPE: 12.5023\tvalid_1's l1: 0.128111\tvalid_1's SMAPE: 13.1543\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "13.154291115667144"
            ]
          },
          "metadata": {},
          "execution_count": 26
        }
      ],
      "source": [
        "lgbtrain = lgb.Dataset(data=X_train, label=Y_train, feature_name=cols)\n",
        "lgbval = lgb.Dataset(data=X_val, label=Y_val, reference=lgbtrain, feature_name=cols)\n",
        "\n",
        "model = lgb.train(lgb_params, lgbtrain,\n",
        "                  valid_sets=[lgbtrain, lgbval],\n",
        "                  num_boost_round=lgb_params['num_boost_round'],\n",
        "                  early_stopping_rounds=lgb_params['early_stopping_rounds'],\n",
        "                  feval=lgbm_smape, \n",
        "                  verbose_eval=100)\n",
        "\n",
        "y_pred_val = model.predict(X_val, num_iteration=model.best_iteration)\n",
        "\n",
        "# percentage of validation error\n",
        "smape(np.expm1(y_pred_val), np.expm1(Y_val))"
      ],
      "id": "f7890a9d"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0e142746"
      },
      "source": [
        "### Final Model"
      ],
      "id": "0e142746"
    },
    {
      "cell_type": "code",
      "execution_count": 27,
      "metadata": {
        "id": "55d63960"
      },
      "outputs": [],
      "source": [
        "# determination of test and train dependent/independent variables\n",
        "\n",
        "train = df.loc[~df.sales.isna()]\n",
        "Y_train = train['sales']\n",
        "X_train = train[cols]\n",
        "\n",
        "test = df.loc[df.sales.isna()]\n",
        "X_test = test[cols]"
      ],
      "id": "55d63960"
    },
    {
      "cell_type": "code",
      "execution_count": 28,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "43e6c24a",
        "outputId": "b62ecaa5-f076-42db-e940-f906ab574543"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[LightGBM] [Warning] Auto-choosing col-wise multi-threading, the overhead of testing was 0.436173 seconds.\n",
            "You can set `force_col_wise=true` to remove the overhead.\n"
          ]
        }
      ],
      "source": [
        "lgb_params = {'metric': {'mae'},\n",
        "              'num_leaves': 10,\n",
        "              'learning_rate': 0.02,\n",
        "              'feature_fraction': 0.8,\n",
        "              'max_depth': 5,\n",
        "              'verbose': 0,\n",
        "              'nthread': -1,\n",
        "              \"num_boost_round\": model.best_iteration}\n",
        "\n",
        "# LightGBM dataset\n",
        "lgbtrain_all = lgb.Dataset(data=X_train, label=Y_train, feature_name=cols)\n",
        "\n",
        "model = lgb.train(lgb_params, lgbtrain_all, num_boost_round=model.best_iteration)\n",
        "test_preds = model.predict(X_test, num_iteration=model.best_iteration)"
      ],
      "id": "43e6c24a"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "70823920"
      },
      "source": [
        "## 8. Time Series Analysis\n",
        "\n",
        "* In this section, first of all, the train data set will be reduced to a weekly basis.\n",
        "* Then, using the weekly data set, respectively:\n",
        "* LightGBM Model\n",
        "* Single Exponential Smoothing\n",
        "* Double Exponential Smoothing\n",
        "* Triple Exponential Smoothing\n",
        "* ARIMA\n",
        "* Sales demand forecasting models for 2017 will be created with SARIMA.\n",
        "* Actual values ​​will be compared with estimated values.\n"
      ],
      "id": "70823920"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cf13bd9d"
      },
      "source": [
        "### Load Dataset and EDA"
      ],
      "id": "cf13bd9d"
    },
    {
      "cell_type": "code",
      "execution_count": 31,
      "metadata": {
        "id": "688ad553",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "outputId": "0c69f5d9-e0a6-436c-d49d-026725b984fb"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "        date  store  item      sales\n",
              "0 2013-01-06    5.5  25.5  30.355667\n",
              "1 2013-01-13    5.5  25.5  29.686286\n",
              "2 2013-01-20    5.5  25.5  29.414286\n",
              "3 2013-01-27    5.5  25.5  29.513714\n",
              "4 2013-02-03    5.5  25.5  31.150571"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-15980bf4-9122-4ea2-bcbc-0e7757bd482b\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>date</th>\n",
              "      <th>store</th>\n",
              "      <th>item</th>\n",
              "      <th>sales</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>2013-01-06</td>\n",
              "      <td>5.5</td>\n",
              "      <td>25.5</td>\n",
              "      <td>30.355667</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>2013-01-13</td>\n",
              "      <td>5.5</td>\n",
              "      <td>25.5</td>\n",
              "      <td>29.686286</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2013-01-20</td>\n",
              "      <td>5.5</td>\n",
              "      <td>25.5</td>\n",
              "      <td>29.414286</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>2013-01-27</td>\n",
              "      <td>5.5</td>\n",
              "      <td>25.5</td>\n",
              "      <td>29.513714</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>2013-02-03</td>\n",
              "      <td>5.5</td>\n",
              "      <td>25.5</td>\n",
              "      <td>31.150571</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-15980bf4-9122-4ea2-bcbc-0e7757bd482b')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-15980bf4-9122-4ea2-bcbc-0e7757bd482b button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-15980bf4-9122-4ea2-bcbc-0e7757bd482b');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 31
        }
      ],
      "source": [
        "data = pd.read_csv('train.csv', parse_dates=['date'])\n",
        "data.head()\n",
        "data.shape\n",
        "\n",
        "# reduce dataset to weekly\n",
        "data.set_index(\"date\",inplace=True)\n",
        "df= data.resample(\"W\").mean()\n",
        "df.reset_index(inplace=True)\n",
        "df.head()\n",
        "df.shape\n",
        "\n",
        "df.index.freq = \"W\"\n",
        "df.head()"
      ],
      "id": "688ad553"
    },
    {
      "cell_type": "code",
      "execution_count": 32,
      "metadata": {
        "id": "c2489b56",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "fd5fa5b6-04ba-4938-94d6-d9d8f1534373"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(261, 4)"
            ]
          },
          "metadata": {},
          "execution_count": 32
        }
      ],
      "source": [
        "df.shape"
      ],
      "id": "c2489b56"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5be06d52"
      },
      "source": [
        "### Feature Engineering"
      ],
      "id": "5be06d52"
    },
    {
      "cell_type": "code",
      "execution_count": 33,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 357
        },
        "id": "8af6b138",
        "outputId": "b81b5370-9dc6-4c8b-bb2a-7e6fa4c74aa8"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "          date  store  item      sales  month  day_of_month  day_of_year  week_of_year  day_of_week  year  is_wknd  is_month_start  is_month_end  sales_lag_31  sales_lag_61  sales_lag_91  sales_lag_98  sales_lag_105  sales_lag_112  sales_roll_mean_31  sales_roll_mean_61  sales_roll_mean_91  sales_roll_mean_98  sales_roll_mean_105  sales_roll_mean_112  sales_ewm_alpha_099_lag_10  sales_ewm_alpha_099_lag_20  sales_ewm_alpha_099_lag_30  sales_ewm_alpha_099_lag_40  sales_ewm_alpha_099_lag_50  \\\n",
              "256 2017-12-03    5.5  25.5  53.510286     12             3          337            48            6  2017        1               0             0     62.228421     60.280295     48.422658     39.813251      54.423685      50.767521           65.995376           58.859689           60.460285           58.396841            57.257321            59.099008                   61.944343                   75.326579                   66.655854                   44.464902                   42.682299   \n",
              "257 2017-12-10    5.5  25.5  44.114286     12            10          344            49            6  2017        1               0             0     67.371788     56.247940     50.839080     38.487108      42.696507      51.707629           66.866386           61.437072           56.433955           59.773793            56.659792            54.839006                   61.176895                   75.422743                   66.631673                   50.817829                   42.680306   \n",
              "258 2017-12-17    5.5  25.5  44.370286     12            17          351            50            6  2017        1               0             0     67.751113     54.516083     51.247317     38.742207      39.700621      53.873788           66.573825           61.892335           61.422906           57.259506            59.857674            59.207826                   57.585703                   75.263325                   66.745140                   53.133184                   41.904974   \n",
              "259 2017-12-24    5.5  25.5  44.423429     12            24          358            51            6  2017        1               0             0     67.669350     55.061851     51.470565     43.775416      37.193301      53.051733           66.076532           59.659316           57.221574           58.608150            57.262916            56.735838                   57.708474                   67.553308                   66.586743                   53.126072                   40.050730   \n",
              "260 2017-12-31    5.5  25.5  44.417429     12            31          365            52            6  2017        1               0             1     66.655037     55.817378     57.902358     45.276689      40.421110      54.660811           62.640400           62.097231           56.214744           57.256800            58.019688            58.669977                   57.943342                   66.656487                   68.985485                   53.177198                   39.972504   \n",
              "\n",
              "     sales_ewm_alpha_095_lag_10  sales_ewm_alpha_095_lag_20  sales_ewm_alpha_095_lag_30  sales_ewm_alpha_095_lag_40  sales_ewm_alpha_095_lag_50  sales_ewm_alpha_09_lag_10  sales_ewm_alpha_09_lag_20  sales_ewm_alpha_09_lag_30  sales_ewm_alpha_09_lag_40  sales_ewm_alpha_09_lag_50  sales_ewm_alpha_08_lag_10  sales_ewm_alpha_08_lag_20  sales_ewm_alpha_08_lag_30  sales_ewm_alpha_08_lag_40  sales_ewm_alpha_08_lag_50  sales_ewm_alpha_07_lag_10  sales_ewm_alpha_07_lag_20  sales_ewm_alpha_07_lag_30  \\\n",
              "256                   61.949367                   75.308760                   66.468077                   44.460477                   42.696424                  61.957303                  75.273841                  66.233661                  44.454102                  42.750531                  61.987351                  75.156915                  65.763164                  44.431749                  43.010549                  62.055540                  74.967582                  65.279091   \n",
              "257                   61.208154                   75.417967                   66.623261                   50.560924                   42.681093                  61.247959                  75.408727                  66.591652                  50.239210                  42.687310                  61.332784                  75.370354                  66.457776                  49.591950                  42.746338                  61.435062                  75.286875                  66.225727   \n",
              "258                   57.732365                   75.269527                   66.740134                   53.026789                   41.936340                  57.919282                  75.276416                  66.730822                  52.864835                  41.976160                  58.306100                  75.283442                  66.688584                  52.443647                  42.066982                  58.715119                  75.269262                  66.590118   \n",
              "259                   57.710847                   67.865133                   66.592892                   53.121039                   40.127217                  57.730671                  68.255527                  66.599711                  53.099884                  40.226416                  57.828991                  69.037031                  66.605831                  52.989529                  40.438996                  58.011336                  69.813579                  66.586635   \n",
              "260                   57.933971                   66.708314                   68.888873                   53.174881                   39.979489                  57.924210                  66.808238                  68.768714                  53.169931                  39.997184                  57.922370                  67.125349                  68.528938                  53.140077                  40.065171                  57.965401                  67.597274                  68.282791   \n",
              "\n",
              "     sales_ewm_alpha_07_lag_40  sales_ewm_alpha_07_lag_50  sales_ewm_alpha_05_lag_10  sales_ewm_alpha_05_lag_20  sales_ewm_alpha_05_lag_30  sales_ewm_alpha_05_lag_40  sales_ewm_alpha_05_lag_50  \n",
              "256                  44.379961                  43.525509                  62.454476                  74.293534                  64.107567                  44.083539                  45.549427  \n",
              "257                  48.931388                  42.933853                  61.811809                  74.858624                  65.369498                  47.482770                  44.114856  \n",
              "258                  51.889017                  42.208156                  59.680619                  75.060169                  66.057892                  50.319670                  43.005999  \n",
              "259                  52.754905                  40.684847                  58.695167                  71.267799                  66.321517                  51.722835                  41.519000  \n",
              "260                  53.050871                  40.185654                  58.320440                  68.957614                  67.665616                  52.450275                  40.745357  "
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-e399c3f3-0399-46ef-af8c-3e630329b64d\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>date</th>\n",
              "      <th>store</th>\n",
              "      <th>item</th>\n",
              "      <th>sales</th>\n",
              "      <th>month</th>\n",
              "      <th>day_of_month</th>\n",
              "      <th>day_of_year</th>\n",
              "      <th>week_of_year</th>\n",
              "      <th>day_of_week</th>\n",
              "      <th>year</th>\n",
              "      <th>is_wknd</th>\n",
              "      <th>is_month_start</th>\n",
              "      <th>is_month_end</th>\n",
              "      <th>sales_lag_31</th>\n",
              "      <th>sales_lag_61</th>\n",
              "      <th>sales_lag_91</th>\n",
              "      <th>sales_lag_98</th>\n",
              "      <th>sales_lag_105</th>\n",
              "      <th>sales_lag_112</th>\n",
              "      <th>sales_roll_mean_31</th>\n",
              "      <th>sales_roll_mean_61</th>\n",
              "      <th>sales_roll_mean_91</th>\n",
              "      <th>sales_roll_mean_98</th>\n",
              "      <th>sales_roll_mean_105</th>\n",
              "      <th>sales_roll_mean_112</th>\n",
              "      <th>sales_ewm_alpha_099_lag_10</th>\n",
              "      <th>sales_ewm_alpha_099_lag_20</th>\n",
              "      <th>sales_ewm_alpha_099_lag_30</th>\n",
              "      <th>sales_ewm_alpha_099_lag_40</th>\n",
              "      <th>sales_ewm_alpha_099_lag_50</th>\n",
              "      <th>sales_ewm_alpha_095_lag_10</th>\n",
              "      <th>sales_ewm_alpha_095_lag_20</th>\n",
              "      <th>sales_ewm_alpha_095_lag_30</th>\n",
              "      <th>sales_ewm_alpha_095_lag_40</th>\n",
              "      <th>sales_ewm_alpha_095_lag_50</th>\n",
              "      <th>sales_ewm_alpha_09_lag_10</th>\n",
              "      <th>sales_ewm_alpha_09_lag_20</th>\n",
              "      <th>sales_ewm_alpha_09_lag_30</th>\n",
              "      <th>sales_ewm_alpha_09_lag_40</th>\n",
              "      <th>sales_ewm_alpha_09_lag_50</th>\n",
              "      <th>sales_ewm_alpha_08_lag_10</th>\n",
              "      <th>sales_ewm_alpha_08_lag_20</th>\n",
              "      <th>sales_ewm_alpha_08_lag_30</th>\n",
              "      <th>sales_ewm_alpha_08_lag_40</th>\n",
              "      <th>sales_ewm_alpha_08_lag_50</th>\n",
              "      <th>sales_ewm_alpha_07_lag_10</th>\n",
              "      <th>sales_ewm_alpha_07_lag_20</th>\n",
              "      <th>sales_ewm_alpha_07_lag_30</th>\n",
              "      <th>sales_ewm_alpha_07_lag_40</th>\n",
              "      <th>sales_ewm_alpha_07_lag_50</th>\n",
              "      <th>sales_ewm_alpha_05_lag_10</th>\n",
              "      <th>sales_ewm_alpha_05_lag_20</th>\n",
              "      <th>sales_ewm_alpha_05_lag_30</th>\n",
              "      <th>sales_ewm_alpha_05_lag_40</th>\n",
              "      <th>sales_ewm_alpha_05_lag_50</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>256</th>\n",
              "      <td>2017-12-03</td>\n",
              "      <td>5.5</td>\n",
              "      <td>25.5</td>\n",
              "      <td>53.510286</td>\n",
              "      <td>12</td>\n",
              "      <td>3</td>\n",
              "      <td>337</td>\n",
              "      <td>48</td>\n",
              "      <td>6</td>\n",
              "      <td>2017</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>62.228421</td>\n",
              "      <td>60.280295</td>\n",
              "      <td>48.422658</td>\n",
              "      <td>39.813251</td>\n",
              "      <td>54.423685</td>\n",
              "      <td>50.767521</td>\n",
              "      <td>65.995376</td>\n",
              "      <td>58.859689</td>\n",
              "      <td>60.460285</td>\n",
              "      <td>58.396841</td>\n",
              "      <td>57.257321</td>\n",
              "      <td>59.099008</td>\n",
              "      <td>61.944343</td>\n",
              "      <td>75.326579</td>\n",
              "      <td>66.655854</td>\n",
              "      <td>44.464902</td>\n",
              "      <td>42.682299</td>\n",
              "      <td>61.949367</td>\n",
              "      <td>75.308760</td>\n",
              "      <td>66.468077</td>\n",
              "      <td>44.460477</td>\n",
              "      <td>42.696424</td>\n",
              "      <td>61.957303</td>\n",
              "      <td>75.273841</td>\n",
              "      <td>66.233661</td>\n",
              "      <td>44.454102</td>\n",
              "      <td>42.750531</td>\n",
              "      <td>61.987351</td>\n",
              "      <td>75.156915</td>\n",
              "      <td>65.763164</td>\n",
              "      <td>44.431749</td>\n",
              "      <td>43.010549</td>\n",
              "      <td>62.055540</td>\n",
              "      <td>74.967582</td>\n",
              "      <td>65.279091</td>\n",
              "      <td>44.379961</td>\n",
              "      <td>43.525509</td>\n",
              "      <td>62.454476</td>\n",
              "      <td>74.293534</td>\n",
              "      <td>64.107567</td>\n",
              "      <td>44.083539</td>\n",
              "      <td>45.549427</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>257</th>\n",
              "      <td>2017-12-10</td>\n",
              "      <td>5.5</td>\n",
              "      <td>25.5</td>\n",
              "      <td>44.114286</td>\n",
              "      <td>12</td>\n",
              "      <td>10</td>\n",
              "      <td>344</td>\n",
              "      <td>49</td>\n",
              "      <td>6</td>\n",
              "      <td>2017</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>67.371788</td>\n",
              "      <td>56.247940</td>\n",
              "      <td>50.839080</td>\n",
              "      <td>38.487108</td>\n",
              "      <td>42.696507</td>\n",
              "      <td>51.707629</td>\n",
              "      <td>66.866386</td>\n",
              "      <td>61.437072</td>\n",
              "      <td>56.433955</td>\n",
              "      <td>59.773793</td>\n",
              "      <td>56.659792</td>\n",
              "      <td>54.839006</td>\n",
              "      <td>61.176895</td>\n",
              "      <td>75.422743</td>\n",
              "      <td>66.631673</td>\n",
              "      <td>50.817829</td>\n",
              "      <td>42.680306</td>\n",
              "      <td>61.208154</td>\n",
              "      <td>75.417967</td>\n",
              "      <td>66.623261</td>\n",
              "      <td>50.560924</td>\n",
              "      <td>42.681093</td>\n",
              "      <td>61.247959</td>\n",
              "      <td>75.408727</td>\n",
              "      <td>66.591652</td>\n",
              "      <td>50.239210</td>\n",
              "      <td>42.687310</td>\n",
              "      <td>61.332784</td>\n",
              "      <td>75.370354</td>\n",
              "      <td>66.457776</td>\n",
              "      <td>49.591950</td>\n",
              "      <td>42.746338</td>\n",
              "      <td>61.435062</td>\n",
              "      <td>75.286875</td>\n",
              "      <td>66.225727</td>\n",
              "      <td>48.931388</td>\n",
              "      <td>42.933853</td>\n",
              "      <td>61.811809</td>\n",
              "      <td>74.858624</td>\n",
              "      <td>65.369498</td>\n",
              "      <td>47.482770</td>\n",
              "      <td>44.114856</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>258</th>\n",
              "      <td>2017-12-17</td>\n",
              "      <td>5.5</td>\n",
              "      <td>25.5</td>\n",
              "      <td>44.370286</td>\n",
              "      <td>12</td>\n",
              "      <td>17</td>\n",
              "      <td>351</td>\n",
              "      <td>50</td>\n",
              "      <td>6</td>\n",
              "      <td>2017</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>67.751113</td>\n",
              "      <td>54.516083</td>\n",
              "      <td>51.247317</td>\n",
              "      <td>38.742207</td>\n",
              "      <td>39.700621</td>\n",
              "      <td>53.873788</td>\n",
              "      <td>66.573825</td>\n",
              "      <td>61.892335</td>\n",
              "      <td>61.422906</td>\n",
              "      <td>57.259506</td>\n",
              "      <td>59.857674</td>\n",
              "      <td>59.207826</td>\n",
              "      <td>57.585703</td>\n",
              "      <td>75.263325</td>\n",
              "      <td>66.745140</td>\n",
              "      <td>53.133184</td>\n",
              "      <td>41.904974</td>\n",
              "      <td>57.732365</td>\n",
              "      <td>75.269527</td>\n",
              "      <td>66.740134</td>\n",
              "      <td>53.026789</td>\n",
              "      <td>41.936340</td>\n",
              "      <td>57.919282</td>\n",
              "      <td>75.276416</td>\n",
              "      <td>66.730822</td>\n",
              "      <td>52.864835</td>\n",
              "      <td>41.976160</td>\n",
              "      <td>58.306100</td>\n",
              "      <td>75.283442</td>\n",
              "      <td>66.688584</td>\n",
              "      <td>52.443647</td>\n",
              "      <td>42.066982</td>\n",
              "      <td>58.715119</td>\n",
              "      <td>75.269262</td>\n",
              "      <td>66.590118</td>\n",
              "      <td>51.889017</td>\n",
              "      <td>42.208156</td>\n",
              "      <td>59.680619</td>\n",
              "      <td>75.060169</td>\n",
              "      <td>66.057892</td>\n",
              "      <td>50.319670</td>\n",
              "      <td>43.005999</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>259</th>\n",
              "      <td>2017-12-24</td>\n",
              "      <td>5.5</td>\n",
              "      <td>25.5</td>\n",
              "      <td>44.423429</td>\n",
              "      <td>12</td>\n",
              "      <td>24</td>\n",
              "      <td>358</td>\n",
              "      <td>51</td>\n",
              "      <td>6</td>\n",
              "      <td>2017</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>67.669350</td>\n",
              "      <td>55.061851</td>\n",
              "      <td>51.470565</td>\n",
              "      <td>43.775416</td>\n",
              "      <td>37.193301</td>\n",
              "      <td>53.051733</td>\n",
              "      <td>66.076532</td>\n",
              "      <td>59.659316</td>\n",
              "      <td>57.221574</td>\n",
              "      <td>58.608150</td>\n",
              "      <td>57.262916</td>\n",
              "      <td>56.735838</td>\n",
              "      <td>57.708474</td>\n",
              "      <td>67.553308</td>\n",
              "      <td>66.586743</td>\n",
              "      <td>53.126072</td>\n",
              "      <td>40.050730</td>\n",
              "      <td>57.710847</td>\n",
              "      <td>67.865133</td>\n",
              "      <td>66.592892</td>\n",
              "      <td>53.121039</td>\n",
              "      <td>40.127217</td>\n",
              "      <td>57.730671</td>\n",
              "      <td>68.255527</td>\n",
              "      <td>66.599711</td>\n",
              "      <td>53.099884</td>\n",
              "      <td>40.226416</td>\n",
              "      <td>57.828991</td>\n",
              "      <td>69.037031</td>\n",
              "      <td>66.605831</td>\n",
              "      <td>52.989529</td>\n",
              "      <td>40.438996</td>\n",
              "      <td>58.011336</td>\n",
              "      <td>69.813579</td>\n",
              "      <td>66.586635</td>\n",
              "      <td>52.754905</td>\n",
              "      <td>40.684847</td>\n",
              "      <td>58.695167</td>\n",
              "      <td>71.267799</td>\n",
              "      <td>66.321517</td>\n",
              "      <td>51.722835</td>\n",
              "      <td>41.519000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>260</th>\n",
              "      <td>2017-12-31</td>\n",
              "      <td>5.5</td>\n",
              "      <td>25.5</td>\n",
              "      <td>44.417429</td>\n",
              "      <td>12</td>\n",
              "      <td>31</td>\n",
              "      <td>365</td>\n",
              "      <td>52</td>\n",
              "      <td>6</td>\n",
              "      <td>2017</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>66.655037</td>\n",
              "      <td>55.817378</td>\n",
              "      <td>57.902358</td>\n",
              "      <td>45.276689</td>\n",
              "      <td>40.421110</td>\n",
              "      <td>54.660811</td>\n",
              "      <td>62.640400</td>\n",
              "      <td>62.097231</td>\n",
              "      <td>56.214744</td>\n",
              "      <td>57.256800</td>\n",
              "      <td>58.019688</td>\n",
              "      <td>58.669977</td>\n",
              "      <td>57.943342</td>\n",
              "      <td>66.656487</td>\n",
              "      <td>68.985485</td>\n",
              "      <td>53.177198</td>\n",
              "      <td>39.972504</td>\n",
              "      <td>57.933971</td>\n",
              "      <td>66.708314</td>\n",
              "      <td>68.888873</td>\n",
              "      <td>53.174881</td>\n",
              "      <td>39.979489</td>\n",
              "      <td>57.924210</td>\n",
              "      <td>66.808238</td>\n",
              "      <td>68.768714</td>\n",
              "      <td>53.169931</td>\n",
              "      <td>39.997184</td>\n",
              "      <td>57.922370</td>\n",
              "      <td>67.125349</td>\n",
              "      <td>68.528938</td>\n",
              "      <td>53.140077</td>\n",
              "      <td>40.065171</td>\n",
              "      <td>57.965401</td>\n",
              "      <td>67.597274</td>\n",
              "      <td>68.282791</td>\n",
              "      <td>53.050871</td>\n",
              "      <td>40.185654</td>\n",
              "      <td>58.320440</td>\n",
              "      <td>68.957614</td>\n",
              "      <td>67.665616</td>\n",
              "      <td>52.450275</td>\n",
              "      <td>40.745357</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-e399c3f3-0399-46ef-af8c-3e630329b64d')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-e399c3f3-0399-46ef-af8c-3e630329b64d button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-e399c3f3-0399-46ef-af8c-3e630329b64d');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 33
        }
      ],
      "source": [
        "# Month\n",
        "df['month'] = df.date.dt.month\n",
        "# Day of Month\n",
        "df['day_of_month'] = df.date.dt.day\n",
        "# Day of year\n",
        "df['day_of_year'] = df.date.dt.dayofyear\n",
        "# Week of year\n",
        "df['week_of_year'] = df.date.dt.weekofyear\n",
        "# Day of week\n",
        "df['day_of_week'] = df.date.dt.dayofweek\n",
        "# Year\n",
        "df['year'] = df.date.dt.year\n",
        "# Weekend check\n",
        "df[\"is_wknd\"] = df.date.dt.weekday // 4\n",
        "# Month start check\n",
        "df['is_month_start'] = df.date.dt.is_month_start.astype(int)\n",
        "# Month end check\n",
        "df['is_month_end'] = df.date.dt.is_month_end.astype(int)\n",
        "\n",
        "# Lag/Shifted Features (Delays)\n",
        "def lag_features(dataframe, lags):\n",
        "    for lag in lags:\n",
        "        dataframe['sales_lag_' + str(lag)] = dataframe['sales'].transform(\n",
        "            lambda x: x.shift(lag)) + random_noise(dataframe)\n",
        "    return dataframe\n",
        "\n",
        "df = lag_features(df, [31, 61, 91, 98, 105, 112])\n",
        "\n",
        "\n",
        "# Moving Average Features\n",
        "def roll_mean_features(dataframe, windows):\n",
        "    for window in windows:\n",
        "        dataframe['sales_roll_mean_' + str(window)] = dataframe['sales']. \\\n",
        "                                                          transform(\n",
        "            lambda x: x.shift(1).rolling(window=window, min_periods=10, win_type=\"triang\").mean()) + random_noise(\n",
        "            dataframe)\n",
        "    return dataframe\n",
        "\n",
        "\n",
        "df = roll_mean_features(df, [31, 61, 91, 98, 105, 112])\n",
        "\n",
        "\n",
        "# Exponentially Weighted Mean Features\n",
        "def ewm_features(dataframe, alphas, lags):\n",
        "    for alpha in alphas:\n",
        "        for lag in lags:\n",
        "            dataframe['sales_ewm_alpha_' + str(alpha).replace(\".\", \"\") + \"_lag_\" + str(lag)] = \\\n",
        "                dataframe['sales'].transform(lambda x: x.shift(lag).ewm(alpha=alpha).mean())\n",
        "    return dataframe\n",
        "\n",
        "\n",
        "alphas = [0.99, 0.95, 0.9, 0.8, 0.7, 0.5]\n",
        "lags = [10, 20, 30, 40, 50]\n",
        "\n",
        "df = ewm_features(df, alphas, lags)\n",
        "\n",
        "df.tail()\n"
      ],
      "id": "8af6b138"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "548d858a"
      },
      "source": [
        "### LightGBM Model"
      ],
      "id": "548d858a"
    },
    {
      "cell_type": "code",
      "execution_count": 34,
      "metadata": {
        "id": "7a31c8c1"
      },
      "outputs": [],
      "source": [
        "\n",
        "#df = pd.get_dummies(df, columns=['day_of_week', 'month'])\n",
        "\n",
        "df = pd.get_dummies(df)\n",
        "\n",
        "# Converting sales to log(1+sales)\n",
        "df['sales'] = np.log1p(df[\"sales\"].values)\n",
        "\n",
        "# train-test data selection\n",
        "train = df.loc[(df[\"date\"] < \"2017-01-01\"), :]\n",
        "test = df.loc[(df[\"date\"] >= \"2017-01-01\"), :]\n",
        "\n",
        "# Dependent and Independent variables\n",
        "cols = [col for col in train.columns if col not in ['date', \"sales\", \"year\"]]\n",
        "X_train = train[cols]\n",
        "Y_train = train['sales']\n",
        "X_test = test[cols]\n",
        "Y_test = test[\"sales\"]\n"
      ],
      "id": "7a31c8c1"
    },
    {
      "cell_type": "code",
      "execution_count": 35,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "45df8046",
        "outputId": "5aca57f3-78cb-4ed5-b55f-e8b9e810b45f"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[LightGBM] [Warning] Unknown parameter: feature_name\n",
            "[LightGBM] [Warning] Unknown parameter: feature_name\n",
            "[LightGBM] [Warning] Auto-choosing col-wise multi-threading, the overhead of testing was 0.000126 seconds.\n",
            "You can set `force_col_wise=true` to remove the overhead.\n",
            "[LightGBM] [Warning] Unknown parameter: feature_name\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1]\ttraining's l1: 0.172041\tvalid_1's l1: 0.213187\n",
            "Training until validation scores don't improve for 200 rounds\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[2]\ttraining's l1: 0.169049\tvalid_1's l1: 0.208704\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[3]\ttraining's l1: 0.16612\tvalid_1's l1: 0.204311\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[4]\ttraining's l1: 0.163298\tvalid_1's l1: 0.200006\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[5]\ttraining's l1: 0.160566\tvalid_1's l1: 0.195748\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[6]\ttraining's l1: 0.157849\tvalid_1's l1: 0.191627\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[7]\ttraining's l1: 0.155244\tvalid_1's l1: 0.187538\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[8]\ttraining's l1: 0.152661\tvalid_1's l1: 0.183672\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[9]\ttraining's l1: 0.150164\tvalid_1's l1: 0.179786\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[10]\ttraining's l1: 0.147707\tvalid_1's l1: 0.175933\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[11]\ttraining's l1: 0.14531\tvalid_1's l1: 0.172287\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[12]\ttraining's l1: 0.142939\tvalid_1's l1: 0.16881\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[13]\ttraining's l1: 0.140686\tvalid_1's l1: 0.165292\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[14]\ttraining's l1: 0.13838\tvalid_1's l1: 0.162015\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[15]\ttraining's l1: 0.13616\tvalid_1's l1: 0.158511\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[16]\ttraining's l1: 0.133956\tvalid_1's l1: 0.155139\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[17]\ttraining's l1: 0.131685\tvalid_1's l1: 0.152204\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[18]\ttraining's l1: 0.129645\tvalid_1's l1: 0.149\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[19]\ttraining's l1: 0.12755\tvalid_1's l1: 0.146129\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[20]\ttraining's l1: 0.125547\tvalid_1's l1: 0.142972\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[21]\ttraining's l1: 0.123545\tvalid_1's l1: 0.14041\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[22]\ttraining's l1: 0.121507\tvalid_1's l1: 0.138048\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[23]\ttraining's l1: 0.119485\tvalid_1's l1: 0.135416\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[24]\ttraining's l1: 0.117503\tvalid_1's l1: 0.132922\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[25]\ttraining's l1: 0.11573\tvalid_1's l1: 0.130211\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[26]\ttraining's l1: 0.113844\tvalid_1's l1: 0.128101\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[27]\ttraining's l1: 0.111993\tvalid_1's l1: 0.126016\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[28]\ttraining's l1: 0.110185\tvalid_1's l1: 0.123973\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[29]\ttraining's l1: 0.108406\tvalid_1's l1: 0.122025\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[30]\ttraining's l1: 0.10666\tvalid_1's l1: 0.119833\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[31]\ttraining's l1: 0.104975\tvalid_1's l1: 0.117798\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[32]\ttraining's l1: 0.103335\tvalid_1's l1: 0.115456\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[33]\ttraining's l1: 0.101743\tvalid_1's l1: 0.113595\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[34]\ttraining's l1: 0.100142\tvalid_1's l1: 0.111497\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[35]\ttraining's l1: 0.0985622\tvalid_1's l1: 0.109528\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[36]\ttraining's l1: 0.0970525\tvalid_1's l1: 0.107753\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[37]\ttraining's l1: 0.0955366\tvalid_1's l1: 0.105679\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[38]\ttraining's l1: 0.0940568\tvalid_1's l1: 0.104\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[39]\ttraining's l1: 0.09266\tvalid_1's l1: 0.10232\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[40]\ttraining's l1: 0.0912908\tvalid_1's l1: 0.100409\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[41]\ttraining's l1: 0.0899113\tvalid_1's l1: 0.0987973\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[42]\ttraining's l1: 0.0885965\tvalid_1's l1: 0.0969695\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[43]\ttraining's l1: 0.0873353\tvalid_1's l1: 0.0953606\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[44]\ttraining's l1: 0.0860574\tvalid_1's l1: 0.0935423\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[45]\ttraining's l1: 0.0847328\tvalid_1's l1: 0.0921037\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[46]\ttraining's l1: 0.0835328\tvalid_1's l1: 0.0903931\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[47]\ttraining's l1: 0.0822952\tvalid_1's l1: 0.0888099\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[48]\ttraining's l1: 0.081132\tvalid_1's l1: 0.0874152\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[49]\ttraining's l1: 0.0799576\tvalid_1's l1: 0.0862938\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[50]\ttraining's l1: 0.0788338\tvalid_1's l1: 0.0848547\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[51]\ttraining's l1: 0.0777601\tvalid_1's l1: 0.0837912\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[52]\ttraining's l1: 0.0767396\tvalid_1's l1: 0.0825322\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[53]\ttraining's l1: 0.0756814\tvalid_1's l1: 0.0812364\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[54]\ttraining's l1: 0.0745961\tvalid_1's l1: 0.0799114\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[55]\ttraining's l1: 0.0735347\tvalid_1's l1: 0.0785949\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[56]\ttraining's l1: 0.0724103\tvalid_1's l1: 0.0774184\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[57]\ttraining's l1: 0.0714489\tvalid_1's l1: 0.0762246\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[58]\ttraining's l1: 0.0704582\tvalid_1's l1: 0.0750818\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[59]\ttraining's l1: 0.0694185\tvalid_1's l1: 0.0740294\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[60]\ttraining's l1: 0.068437\tvalid_1's l1: 0.0730487\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[61]\ttraining's l1: 0.0675559\tvalid_1's l1: 0.0720411\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[62]\ttraining's l1: 0.0665906\tvalid_1's l1: 0.0709922\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[63]\ttraining's l1: 0.0656871\tvalid_1's l1: 0.0698897\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[64]\ttraining's l1: 0.0649033\tvalid_1's l1: 0.0687453\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[65]\ttraining's l1: 0.0640295\tvalid_1's l1: 0.0678508\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[66]\ttraining's l1: 0.0631456\tvalid_1's l1: 0.0667784\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[67]\ttraining's l1: 0.0622564\tvalid_1's l1: 0.0658936\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[68]\ttraining's l1: 0.0614375\tvalid_1's l1: 0.0648343\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[69]\ttraining's l1: 0.0605527\tvalid_1's l1: 0.0640336\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[70]\ttraining's l1: 0.0597422\tvalid_1's l1: 0.0632182\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[71]\ttraining's l1: 0.0588944\tvalid_1's l1: 0.0625081\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[72]\ttraining's l1: 0.0580743\tvalid_1's l1: 0.0617834\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[73]\ttraining's l1: 0.0573055\tvalid_1's l1: 0.0612219\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[74]\ttraining's l1: 0.0565861\tvalid_1's l1: 0.0604144\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[75]\ttraining's l1: 0.0558408\tvalid_1's l1: 0.0598318\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[76]\ttraining's l1: 0.0551281\tvalid_1's l1: 0.0593386\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[77]\ttraining's l1: 0.0543848\tvalid_1's l1: 0.058797\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[78]\ttraining's l1: 0.0537005\tvalid_1's l1: 0.0582857\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[79]\ttraining's l1: 0.0530531\tvalid_1's l1: 0.0576759\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[80]\ttraining's l1: 0.052377\tvalid_1's l1: 0.0572112\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[81]\ttraining's l1: 0.0517513\tvalid_1's l1: 0.0567297\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[82]\ttraining's l1: 0.0511372\tvalid_1's l1: 0.0562562\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[83]\ttraining's l1: 0.0504514\tvalid_1's l1: 0.055876\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[84]\ttraining's l1: 0.0498331\tvalid_1's l1: 0.0556407\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[85]\ttraining's l1: 0.0492195\tvalid_1's l1: 0.0553945\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[86]\ttraining's l1: 0.0486317\tvalid_1's l1: 0.0551533\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[87]\ttraining's l1: 0.0480433\tvalid_1's l1: 0.0549045\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[88]\ttraining's l1: 0.0475118\tvalid_1's l1: 0.0545411\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[89]\ttraining's l1: 0.0469514\tvalid_1's l1: 0.0543033\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[90]\ttraining's l1: 0.0463817\tvalid_1's l1: 0.0538242\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[91]\ttraining's l1: 0.0458807\tvalid_1's l1: 0.0536732\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[92]\ttraining's l1: 0.0453919\tvalid_1's l1: 0.0534815\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[93]\ttraining's l1: 0.0449224\tvalid_1's l1: 0.0533102\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[94]\ttraining's l1: 0.0444714\tvalid_1's l1: 0.0530384\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[95]\ttraining's l1: 0.0439965\tvalid_1's l1: 0.0526395\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[96]\ttraining's l1: 0.0435136\tvalid_1's l1: 0.0523047\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[97]\ttraining's l1: 0.0430632\tvalid_1's l1: 0.0521201\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[98]\ttraining's l1: 0.0426222\tvalid_1's l1: 0.0517833\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[99]\ttraining's l1: 0.042177\tvalid_1's l1: 0.0515231\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[100]\ttraining's l1: 0.0417935\tvalid_1's l1: 0.051195\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[101]\ttraining's l1: 0.0413699\tvalid_1's l1: 0.0509036\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[102]\ttraining's l1: 0.0409844\tvalid_1's l1: 0.0506181\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[103]\ttraining's l1: 0.0406001\tvalid_1's l1: 0.0504546\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[104]\ttraining's l1: 0.0402591\tvalid_1's l1: 0.0501594\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[105]\ttraining's l1: 0.039901\tvalid_1's l1: 0.0499399\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[106]\ttraining's l1: 0.0395847\tvalid_1's l1: 0.0496638\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[107]\ttraining's l1: 0.0392142\tvalid_1's l1: 0.0494136\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[108]\ttraining's l1: 0.0388839\tvalid_1's l1: 0.0491472\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[109]\ttraining's l1: 0.0385447\tvalid_1's l1: 0.0488845\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[110]\ttraining's l1: 0.0382193\tvalid_1's l1: 0.0486476\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[111]\ttraining's l1: 0.0379161\tvalid_1's l1: 0.0484471\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[112]\ttraining's l1: 0.0376252\tvalid_1's l1: 0.0482229\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[113]\ttraining's l1: 0.0373197\tvalid_1's l1: 0.0479791\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[114]\ttraining's l1: 0.0370069\tvalid_1's l1: 0.0478357\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[115]\ttraining's l1: 0.0367394\tvalid_1's l1: 0.0476571\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[116]\ttraining's l1: 0.0364779\tvalid_1's l1: 0.0474755\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[117]\ttraining's l1: 0.0361993\tvalid_1's l1: 0.0473569\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[118]\ttraining's l1: 0.0359261\tvalid_1's l1: 0.0471693\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[119]\ttraining's l1: 0.0356576\tvalid_1's l1: 0.0469553\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[120]\ttraining's l1: 0.0354073\tvalid_1's l1: 0.0468516\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[121]\ttraining's l1: 0.0351448\tvalid_1's l1: 0.04667\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[122]\ttraining's l1: 0.0349364\tvalid_1's l1: 0.0464891\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[123]\ttraining's l1: 0.0347004\tvalid_1's l1: 0.0463875\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[124]\ttraining's l1: 0.0344716\tvalid_1's l1: 0.0462879\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[125]\ttraining's l1: 0.0342527\tvalid_1's l1: 0.046174\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[126]\ttraining's l1: 0.0339762\tvalid_1's l1: 0.0459439\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[127]\ttraining's l1: 0.0337335\tvalid_1's l1: 0.0457592\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[128]\ttraining's l1: 0.0334728\tvalid_1's l1: 0.0455362\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[129]\ttraining's l1: 0.033297\tvalid_1's l1: 0.0453876\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[130]\ttraining's l1: 0.0330661\tvalid_1's l1: 0.0451969\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[131]\ttraining's l1: 0.0329108\tvalid_1's l1: 0.0450522\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[132]\ttraining's l1: 0.0326662\tvalid_1's l1: 0.0449064\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[133]\ttraining's l1: 0.032423\tvalid_1's l1: 0.0447347\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[134]\ttraining's l1: 0.0322445\tvalid_1's l1: 0.0446318\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[135]\ttraining's l1: 0.0320107\tvalid_1's l1: 0.0444853\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[136]\ttraining's l1: 0.0317989\tvalid_1's l1: 0.0443535\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[137]\ttraining's l1: 0.0315927\tvalid_1's l1: 0.0442439\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[138]\ttraining's l1: 0.0314533\tvalid_1's l1: 0.0441772\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[139]\ttraining's l1: 0.0312582\tvalid_1's l1: 0.044014\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[140]\ttraining's l1: 0.0310238\tvalid_1's l1: 0.043832\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[141]\ttraining's l1: 0.0308108\tvalid_1's l1: 0.0436768\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[142]\ttraining's l1: 0.0306315\tvalid_1's l1: 0.0436181\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[143]\ttraining's l1: 0.0303861\tvalid_1's l1: 0.0435189\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[144]\ttraining's l1: 0.0302239\tvalid_1's l1: 0.0433709\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[145]\ttraining's l1: 0.0301034\tvalid_1's l1: 0.0432989\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[146]\ttraining's l1: 0.029947\tvalid_1's l1: 0.0431621\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[147]\ttraining's l1: 0.0297335\tvalid_1's l1: 0.0430316\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[148]\ttraining's l1: 0.0296202\tvalid_1's l1: 0.0429546\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[149]\ttraining's l1: 0.0294556\tvalid_1's l1: 0.0429309\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[150]\ttraining's l1: 0.0293449\tvalid_1's l1: 0.0428555\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[151]\ttraining's l1: 0.0291661\tvalid_1's l1: 0.0427313\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[152]\ttraining's l1: 0.0290018\tvalid_1's l1: 0.0426803\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[153]\ttraining's l1: 0.0288641\tvalid_1's l1: 0.0425597\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[154]\ttraining's l1: 0.0287213\tvalid_1's l1: 0.0424345\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[155]\ttraining's l1: 0.0286079\tvalid_1's l1: 0.0423634\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[156]\ttraining's l1: 0.0284602\tvalid_1's l1: 0.042262\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[157]\ttraining's l1: 0.0283145\tvalid_1's l1: 0.0421925\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[158]\ttraining's l1: 0.0281778\tvalid_1's l1: 0.0420279\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[159]\ttraining's l1: 0.028066\tvalid_1's l1: 0.0419587\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[160]\ttraining's l1: 0.027912\tvalid_1's l1: 0.0418411\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[161]\ttraining's l1: 0.0277645\tvalid_1's l1: 0.0416981\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[162]\ttraining's l1: 0.0276589\tvalid_1's l1: 0.0416314\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[163]\ttraining's l1: 0.0275219\tvalid_1's l1: 0.0415366\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[164]\ttraining's l1: 0.0274106\tvalid_1's l1: 0.0414241\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[165]\ttraining's l1: 0.0272645\tvalid_1's l1: 0.0413085\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[166]\ttraining's l1: 0.0271328\tvalid_1's l1: 0.0412118\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[167]\ttraining's l1: 0.0270373\tvalid_1's l1: 0.0410769\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[168]\ttraining's l1: 0.0269423\tvalid_1's l1: 0.0410027\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[169]\ttraining's l1: 0.0268401\tvalid_1's l1: 0.0409274\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[170]\ttraining's l1: 0.026735\tvalid_1's l1: 0.0408503\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[171]\ttraining's l1: 0.0265969\tvalid_1's l1: 0.0407552\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[172]\ttraining's l1: 0.0264644\tvalid_1's l1: 0.0406455\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[173]\ttraining's l1: 0.0263749\tvalid_1's l1: 0.0405494\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[174]\ttraining's l1: 0.0262226\tvalid_1's l1: 0.0405018\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[175]\ttraining's l1: 0.0260999\tvalid_1's l1: 0.0403988\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[176]\ttraining's l1: 0.0259862\tvalid_1's l1: 0.0403019\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[177]\ttraining's l1: 0.0258589\tvalid_1's l1: 0.0401316\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[178]\ttraining's l1: 0.0257722\tvalid_1's l1: 0.0400665\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[179]\ttraining's l1: 0.0256301\tvalid_1's l1: 0.0400268\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[180]\ttraining's l1: 0.0254926\tvalid_1's l1: 0.0399878\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[181]\ttraining's l1: 0.0254133\tvalid_1's l1: 0.0398592\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[182]\ttraining's l1: 0.0253255\tvalid_1's l1: 0.0398016\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[183]\ttraining's l1: 0.0251985\tvalid_1's l1: 0.0397765\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[184]\ttraining's l1: 0.0251243\tvalid_1's l1: 0.0396726\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[185]\ttraining's l1: 0.025004\tvalid_1's l1: 0.0396885\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[186]\ttraining's l1: 0.0249275\tvalid_1's l1: 0.0396591\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[187]\ttraining's l1: 0.0248462\tvalid_1's l1: 0.0395365\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[188]\ttraining's l1: 0.0247194\tvalid_1's l1: 0.0394604\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[189]\ttraining's l1: 0.0246476\tvalid_1's l1: 0.0394343\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[190]\ttraining's l1: 0.0245407\tvalid_1's l1: 0.0394077\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[191]\ttraining's l1: 0.0244299\tvalid_1's l1: 0.0394225\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[192]\ttraining's l1: 0.024342\tvalid_1's l1: 0.0394043\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[193]\ttraining's l1: 0.0242493\tvalid_1's l1: 0.0393838\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[194]\ttraining's l1: 0.0241611\tvalid_1's l1: 0.0393261\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[195]\ttraining's l1: 0.0240945\tvalid_1's l1: 0.0393208\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[196]\ttraining's l1: 0.0239816\tvalid_1's l1: 0.0393096\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[197]\ttraining's l1: 0.0238611\tvalid_1's l1: 0.0392384\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[198]\ttraining's l1: 0.0237724\tvalid_1's l1: 0.0392827\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[199]\ttraining's l1: 0.0236996\tvalid_1's l1: 0.0392572\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[200]\ttraining's l1: 0.0236136\tvalid_1's l1: 0.0393145\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[201]\ttraining's l1: 0.0235052\tvalid_1's l1: 0.0393015\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[202]\ttraining's l1: 0.0233955\tvalid_1's l1: 0.0392717\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[203]\ttraining's l1: 0.0233103\tvalid_1's l1: 0.0392444\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[204]\ttraining's l1: 0.0232269\tvalid_1's l1: 0.0392196\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[205]\ttraining's l1: 0.0231513\tvalid_1's l1: 0.0392702\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[206]\ttraining's l1: 0.0230829\tvalid_1's l1: 0.0392386\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[207]\ttraining's l1: 0.0229826\tvalid_1's l1: 0.039225\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[208]\ttraining's l1: 0.0228989\tvalid_1's l1: 0.0391943\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[209]\ttraining's l1: 0.0228286\tvalid_1's l1: 0.0392438\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[210]\ttraining's l1: 0.0227593\tvalid_1's l1: 0.0392245\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[211]\ttraining's l1: 0.0226561\tvalid_1's l1: 0.0391656\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[212]\ttraining's l1: 0.0225589\tvalid_1's l1: 0.0391474\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[213]\ttraining's l1: 0.0224613\tvalid_1's l1: 0.0391248\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[214]\ttraining's l1: 0.0223834\tvalid_1's l1: 0.0390912\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[215]\ttraining's l1: 0.0222926\tvalid_1's l1: 0.0390693\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[216]\ttraining's l1: 0.0222244\tvalid_1's l1: 0.0390386\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[217]\ttraining's l1: 0.0221626\tvalid_1's l1: 0.0389497\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[218]\ttraining's l1: 0.0220806\tvalid_1's l1: 0.0389985\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[219]\ttraining's l1: 0.0220316\tvalid_1's l1: 0.0389432\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[220]\ttraining's l1: 0.0219659\tvalid_1's l1: 0.03892\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[221]\ttraining's l1: 0.0218953\tvalid_1's l1: 0.0389685\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[222]\ttraining's l1: 0.0218466\tvalid_1's l1: 0.0389536\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[223]\ttraining's l1: 0.0217749\tvalid_1's l1: 0.0389644\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[224]\ttraining's l1: 0.0216815\tvalid_1's l1: 0.0389468\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[225]\ttraining's l1: 0.0216195\tvalid_1's l1: 0.0388633\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[226]\ttraining's l1: 0.0215545\tvalid_1's l1: 0.0388416\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[227]\ttraining's l1: 0.0214809\tvalid_1's l1: 0.0389031\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[228]\ttraining's l1: 0.0214195\tvalid_1's l1: 0.0388816\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[229]\ttraining's l1: 0.0213579\tvalid_1's l1: 0.038823\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[230]\ttraining's l1: 0.0213065\tvalid_1's l1: 0.0388397\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[231]\ttraining's l1: 0.0212471\tvalid_1's l1: 0.0387626\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[232]\ttraining's l1: 0.0211633\tvalid_1's l1: 0.0387527\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[233]\ttraining's l1: 0.021104\tvalid_1's l1: 0.0387393\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[234]\ttraining's l1: 0.0210523\tvalid_1's l1: 0.0387552\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[235]\ttraining's l1: 0.0209832\tvalid_1's l1: 0.0387275\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[236]\ttraining's l1: 0.0209364\tvalid_1's l1: 0.0386921\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[237]\ttraining's l1: 0.0208649\tvalid_1's l1: 0.0386725\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[238]\ttraining's l1: 0.0208218\tvalid_1's l1: 0.0386809\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[239]\ttraining's l1: 0.020744\tvalid_1's l1: 0.0386317\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[240]\ttraining's l1: 0.0206998\tvalid_1's l1: 0.0386471\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[241]\ttraining's l1: 0.0206315\tvalid_1's l1: 0.0386159\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[242]\ttraining's l1: 0.0205839\tvalid_1's l1: 0.0385122\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[243]\ttraining's l1: 0.0205347\tvalid_1's l1: 0.0385218\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[244]\ttraining's l1: 0.0204852\tvalid_1's l1: 0.0384646\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[245]\ttraining's l1: 0.0204369\tvalid_1's l1: 0.0384738\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[246]\ttraining's l1: 0.0203823\tvalid_1's l1: 0.0384438\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[247]\ttraining's l1: 0.0203397\tvalid_1's l1: 0.0384725\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[248]\ttraining's l1: 0.0202819\tvalid_1's l1: 0.0384504\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[249]\ttraining's l1: 0.0202245\tvalid_1's l1: 0.0383796\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[250]\ttraining's l1: 0.0201747\tvalid_1's l1: 0.038337\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[251]\ttraining's l1: 0.020133\tvalid_1's l1: 0.0383687\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[252]\ttraining's l1: 0.0200909\tvalid_1's l1: 0.0383696\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[253]\ttraining's l1: 0.020031\tvalid_1's l1: 0.0383361\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[254]\ttraining's l1: 0.0199648\tvalid_1's l1: 0.0382832\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[255]\ttraining's l1: 0.0199234\tvalid_1's l1: 0.0383141\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[256]\ttraining's l1: 0.0198656\tvalid_1's l1: 0.0382816\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[257]\ttraining's l1: 0.0198048\tvalid_1's l1: 0.0382128\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[258]\ttraining's l1: 0.0197605\tvalid_1's l1: 0.0382194\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[259]\ttraining's l1: 0.0197246\tvalid_1's l1: 0.03825\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[260]\ttraining's l1: 0.0196728\tvalid_1's l1: 0.0382281\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[261]\ttraining's l1: 0.0196292\tvalid_1's l1: 0.0382481\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[262]\ttraining's l1: 0.0195886\tvalid_1's l1: 0.0382437\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[263]\ttraining's l1: 0.0195466\tvalid_1's l1: 0.0382583\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[264]\ttraining's l1: 0.0195048\tvalid_1's l1: 0.038211\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[265]\ttraining's l1: 0.0194598\tvalid_1's l1: 0.0381609\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[266]\ttraining's l1: 0.0194162\tvalid_1's l1: 0.0381119\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[267]\ttraining's l1: 0.0193776\tvalid_1's l1: 0.0381077\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[268]\ttraining's l1: 0.0193439\tvalid_1's l1: 0.0381383\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[269]\ttraining's l1: 0.0193112\tvalid_1's l1: 0.0381342\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[270]\ttraining's l1: 0.0192655\tvalid_1's l1: 0.0381201\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[271]\ttraining's l1: 0.0192162\tvalid_1's l1: 0.0380749\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[272]\ttraining's l1: 0.0191777\tvalid_1's l1: 0.0379994\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[273]\ttraining's l1: 0.0191455\tvalid_1's l1: 0.0380292\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[274]\ttraining's l1: 0.0191011\tvalid_1's l1: 0.0380037\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[275]\ttraining's l1: 0.0190536\tvalid_1's l1: 0.0379414\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[276]\ttraining's l1: 0.0190202\tvalid_1's l1: 0.0379677\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[277]\ttraining's l1: 0.0189769\tvalid_1's l1: 0.0379269\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[278]\ttraining's l1: 0.0189162\tvalid_1's l1: 0.0378629\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[279]\ttraining's l1: 0.0188639\tvalid_1's l1: 0.0379249\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[280]\ttraining's l1: 0.0187968\tvalid_1's l1: 0.0378608\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[281]\ttraining's l1: 0.0187558\tvalid_1's l1: 0.0378088\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[282]\ttraining's l1: 0.0187212\tvalid_1's l1: 0.0377236\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[283]\ttraining's l1: 0.0186781\tvalid_1's l1: 0.0377849\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[284]\ttraining's l1: 0.0185983\tvalid_1's l1: 0.0377435\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[285]\ttraining's l1: 0.0185596\tvalid_1's l1: 0.0376787\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[286]\ttraining's l1: 0.0185127\tvalid_1's l1: 0.0376552\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[287]\ttraining's l1: 0.0184868\tvalid_1's l1: 0.037723\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[288]\ttraining's l1: 0.0184446\tvalid_1's l1: 0.0377047\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[289]\ttraining's l1: 0.0184115\tvalid_1's l1: 0.0376924\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[290]\ttraining's l1: 0.0183332\tvalid_1's l1: 0.0376865\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[291]\ttraining's l1: 0.0182691\tvalid_1's l1: 0.0376601\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[292]\ttraining's l1: 0.018221\tvalid_1's l1: 0.0375767\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[293]\ttraining's l1: 0.0181881\tvalid_1's l1: 0.0375861\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[294]\ttraining's l1: 0.0181474\tvalid_1's l1: 0.0375265\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[295]\ttraining's l1: 0.0181036\tvalid_1's l1: 0.0375144\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[296]\ttraining's l1: 0.0180641\tvalid_1's l1: 0.0375025\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[297]\ttraining's l1: 0.0179947\tvalid_1's l1: 0.0374757\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[298]\ttraining's l1: 0.0179306\tvalid_1's l1: 0.0374703\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[299]\ttraining's l1: 0.0178878\tvalid_1's l1: 0.0374589\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[300]\ttraining's l1: 0.0178463\tvalid_1's l1: 0.0373847\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[301]\ttraining's l1: 0.0178088\tvalid_1's l1: 0.0373416\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[302]\ttraining's l1: 0.0177735\tvalid_1's l1: 0.0372819\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[303]\ttraining's l1: 0.0177351\tvalid_1's l1: 0.0372704\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[304]\ttraining's l1: 0.0176762\tvalid_1's l1: 0.0372442\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[305]\ttraining's l1: 0.0176438\tvalid_1's l1: 0.0371792\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[306]\ttraining's l1: 0.0176077\tvalid_1's l1: 0.0371687\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[307]\ttraining's l1: 0.0175752\tvalid_1's l1: 0.0371727\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[308]\ttraining's l1: 0.0175119\tvalid_1's l1: 0.0371677\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[309]\ttraining's l1: 0.0174769\tvalid_1's l1: 0.0371263\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[310]\ttraining's l1: 0.0174203\tvalid_1's l1: 0.0370974\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[311]\ttraining's l1: 0.017356\tvalid_1's l1: 0.0371227\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[312]\ttraining's l1: 0.0173139\tvalid_1's l1: 0.0370903\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[313]\ttraining's l1: 0.01726\tvalid_1's l1: 0.037062\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[314]\ttraining's l1: 0.0172312\tvalid_1's l1: 0.0370039\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[315]\ttraining's l1: 0.0171841\tvalid_1's l1: 0.0369791\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[316]\ttraining's l1: 0.0171384\tvalid_1's l1: 0.0369313\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[317]\ttraining's l1: 0.0171189\tvalid_1's l1: 0.0368736\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[318]\ttraining's l1: 0.0170846\tvalid_1's l1: 0.0368635\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[319]\ttraining's l1: 0.0170208\tvalid_1's l1: 0.0369005\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[320]\ttraining's l1: 0.0169697\tvalid_1's l1: 0.0368963\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[321]\ttraining's l1: 0.0169398\tvalid_1's l1: 0.0368865\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[322]\ttraining's l1: 0.0168913\tvalid_1's l1: 0.0368601\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[323]\ttraining's l1: 0.0168504\tvalid_1's l1: 0.0368294\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[324]\ttraining's l1: 0.0168053\tvalid_1's l1: 0.0368035\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[325]\ttraining's l1: 0.0167771\tvalid_1's l1: 0.0367334\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[326]\ttraining's l1: 0.0167494\tvalid_1's l1: 0.0367706\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[327]\ttraining's l1: 0.0167223\tvalid_1's l1: 0.036716\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[328]\ttraining's l1: 0.0166911\tvalid_1's l1: 0.0367131\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[329]\ttraining's l1: 0.0166587\tvalid_1's l1: 0.036693\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[330]\ttraining's l1: 0.0166371\tvalid_1's l1: 0.0367346\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[331]\ttraining's l1: 0.0165965\tvalid_1's l1: 0.0366865\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[332]\ttraining's l1: 0.0165736\tvalid_1's l1: 0.0367229\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[333]\ttraining's l1: 0.0165425\tvalid_1's l1: 0.0367028\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[334]\ttraining's l1: 0.0165234\tvalid_1's l1: 0.0367383\n",
            "[335]\ttraining's l1: 0.0165019\tvalid_1's l1: 0.0367152\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[336]\ttraining's l1: 0.0164669\tvalid_1's l1: 0.0366624\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[337]\ttraining's l1: 0.0164412\tvalid_1's l1: 0.036652\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[338]\ttraining's l1: 0.0164221\tvalid_1's l1: 0.0366923\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[339]\ttraining's l1: 0.0163784\tvalid_1's l1: 0.0366683\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[340]\ttraining's l1: 0.0163584\tvalid_1's l1: 0.036626\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[341]\ttraining's l1: 0.0163336\tvalid_1's l1: 0.0366219\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[342]\ttraining's l1: 0.0163134\tvalid_1's l1: 0.0366561\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[343]\ttraining's l1: 0.0162816\tvalid_1's l1: 0.0366454\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[344]\ttraining's l1: 0.0162494\tvalid_1's l1: 0.036595\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[345]\ttraining's l1: 0.0162296\tvalid_1's l1: 0.0366335\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[346]\ttraining's l1: 0.0162062\tvalid_1's l1: 0.0366227\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[347]\ttraining's l1: 0.0161797\tvalid_1's l1: 0.0366263\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[348]\ttraining's l1: 0.0161461\tvalid_1's l1: 0.0365891\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[349]\ttraining's l1: 0.0161033\tvalid_1's l1: 0.0365931\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[350]\ttraining's l1: 0.016083\tvalid_1's l1: 0.0365643\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[351]\ttraining's l1: 0.0160578\tvalid_1's l1: 0.0365456\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[352]\ttraining's l1: 0.016035\tvalid_1's l1: 0.0365524\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[353]\ttraining's l1: 0.0160122\tvalid_1's l1: 0.0365419\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[354]\ttraining's l1: 0.0159938\tvalid_1's l1: 0.0365121\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[355]\ttraining's l1: 0.015961\tvalid_1's l1: 0.03648\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[356]\ttraining's l1: 0.0159398\tvalid_1's l1: 0.0364547\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[357]\ttraining's l1: 0.0159178\tvalid_1's l1: 0.0364816\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[358]\ttraining's l1: 0.0158899\tvalid_1's l1: 0.0364584\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[359]\ttraining's l1: 0.0158668\tvalid_1's l1: 0.0364429\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[360]\ttraining's l1: 0.0158323\tvalid_1's l1: 0.0363942\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[361]\ttraining's l1: 0.0158117\tvalid_1's l1: 0.0364008\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[362]\ttraining's l1: 0.0157765\tvalid_1's l1: 0.0363533\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[363]\ttraining's l1: 0.0157564\tvalid_1's l1: 0.0363601\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[364]\ttraining's l1: 0.0157266\tvalid_1's l1: 0.0363285\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[365]\ttraining's l1: 0.0156864\tvalid_1's l1: 0.0362852\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[366]\ttraining's l1: 0.0156647\tvalid_1's l1: 0.036285\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[367]\ttraining's l1: 0.0156489\tvalid_1's l1: 0.0363169\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[368]\ttraining's l1: 0.015628\tvalid_1's l1: 0.0363044\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[369]\ttraining's l1: 0.0156038\tvalid_1's l1: 0.0362696\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[370]\ttraining's l1: 0.0155855\tvalid_1's l1: 0.0362763\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[371]\ttraining's l1: 0.0155568\tvalid_1's l1: 0.0362257\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[372]\ttraining's l1: 0.0155308\tvalid_1's l1: 0.0361729\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[373]\ttraining's l1: 0.0155102\tvalid_1's l1: 0.036208\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[374]\ttraining's l1: 0.0154751\tvalid_1's l1: 0.0361929\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[375]\ttraining's l1: 0.0154527\tvalid_1's l1: 0.0362191\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[376]\ttraining's l1: 0.0154208\tvalid_1's l1: 0.0361747\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[377]\ttraining's l1: 0.015395\tvalid_1's l1: 0.0361612\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[378]\ttraining's l1: 0.0153701\tvalid_1's l1: 0.036113\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[379]\ttraining's l1: 0.0153484\tvalid_1's l1: 0.0360894\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[380]\ttraining's l1: 0.0153211\tvalid_1's l1: 0.0360611\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[381]\ttraining's l1: 0.0152865\tvalid_1's l1: 0.036029\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[382]\ttraining's l1: 0.0152586\tvalid_1's l1: 0.0360143\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[383]\ttraining's l1: 0.0152282\tvalid_1's l1: 0.0360114\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[384]\ttraining's l1: 0.0151896\tvalid_1's l1: 0.0359787\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[385]\ttraining's l1: 0.015172\tvalid_1's l1: 0.0360134\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[386]\ttraining's l1: 0.0151549\tvalid_1's l1: 0.036006\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[387]\ttraining's l1: 0.0151267\tvalid_1's l1: 0.035941\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[388]\ttraining's l1: 0.015108\tvalid_1's l1: 0.0359174\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[389]\ttraining's l1: 0.0150712\tvalid_1's l1: 0.0358679\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[390]\ttraining's l1: 0.0150453\tvalid_1's l1: 0.0358519\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[391]\ttraining's l1: 0.015028\tvalid_1's l1: 0.0358815\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[392]\ttraining's l1: 0.0150097\tvalid_1's l1: 0.0358586\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[393]\ttraining's l1: 0.0149781\tvalid_1's l1: 0.035836\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[394]\ttraining's l1: 0.0149609\tvalid_1's l1: 0.0358289\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[395]\ttraining's l1: 0.0149309\tvalid_1's l1: 0.0357636\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[396]\ttraining's l1: 0.0149045\tvalid_1's l1: 0.0357469\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[397]\ttraining's l1: 0.0148644\tvalid_1's l1: 0.0357159\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[398]\ttraining's l1: 0.0148477\tvalid_1's l1: 0.035719\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[399]\ttraining's l1: 0.014832\tvalid_1's l1: 0.0357166\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[400]\ttraining's l1: 0.0148052\tvalid_1's l1: 0.0356635\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[401]\ttraining's l1: 0.0147828\tvalid_1's l1: 0.0356153\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[402]\ttraining's l1: 0.0147672\tvalid_1's l1: 0.0356373\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[403]\ttraining's l1: 0.0147453\tvalid_1's l1: 0.0355899\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[404]\ttraining's l1: 0.0147124\tvalid_1's l1: 0.0355925\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[405]\ttraining's l1: 0.0146967\tvalid_1's l1: 0.0355957\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[406]\ttraining's l1: 0.014667\tvalid_1's l1: 0.0355905\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[407]\ttraining's l1: 0.0146346\tvalid_1's l1: 0.0355575\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[408]\ttraining's l1: 0.0146021\tvalid_1's l1: 0.0355293\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[409]\ttraining's l1: 0.0145831\tvalid_1's l1: 0.0354919\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[410]\ttraining's l1: 0.0145635\tvalid_1's l1: 0.0355235\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[411]\ttraining's l1: 0.0145345\tvalid_1's l1: 0.0355055\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[412]\ttraining's l1: 0.0144921\tvalid_1's l1: 0.0354561\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[413]\ttraining's l1: 0.0144665\tvalid_1's l1: 0.0354253\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[414]\ttraining's l1: 0.0144516\tvalid_1's l1: 0.0354317\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[415]\ttraining's l1: 0.0144251\tvalid_1's l1: 0.0354\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[416]\ttraining's l1: 0.0144065\tvalid_1's l1: 0.0354315\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[417]\ttraining's l1: 0.0143838\tvalid_1's l1: 0.0353779\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[418]\ttraining's l1: 0.0143473\tvalid_1's l1: 0.0353861\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[419]\ttraining's l1: 0.0143227\tvalid_1's l1: 0.0353393\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[420]\ttraining's l1: 0.0143102\tvalid_1's l1: 0.0353231\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[421]\ttraining's l1: 0.014295\tvalid_1's l1: 0.0353293\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[422]\ttraining's l1: 0.0142776\tvalid_1's l1: 0.0353166\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[423]\ttraining's l1: 0.0142449\tvalid_1's l1: 0.0352958\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[424]\ttraining's l1: 0.0142221\tvalid_1's l1: 0.0352462\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[425]\ttraining's l1: 0.0142042\tvalid_1's l1: 0.0352807\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[426]\ttraining's l1: 0.0141836\tvalid_1's l1: 0.0352365\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[427]\ttraining's l1: 0.0141719\tvalid_1's l1: 0.035245\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[428]\ttraining's l1: 0.0141519\tvalid_1's l1: 0.0351955\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[429]\ttraining's l1: 0.0141341\tvalid_1's l1: 0.0351748\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[430]\ttraining's l1: 0.0141179\tvalid_1's l1: 0.0351618\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[431]\ttraining's l1: 0.0140988\tvalid_1's l1: 0.0351293\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[432]\ttraining's l1: 0.0140714\tvalid_1's l1: 0.0350959\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[433]\ttraining's l1: 0.014045\tvalid_1's l1: 0.0350821\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[434]\ttraining's l1: 0.0140192\tvalid_1's l1: 0.035036\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[435]\ttraining's l1: 0.0139922\tvalid_1's l1: 0.035005\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[436]\ttraining's l1: 0.0139823\tvalid_1's l1: 0.0349939\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[437]\ttraining's l1: 0.0139663\tvalid_1's l1: 0.0350237\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[438]\ttraining's l1: 0.0139402\tvalid_1's l1: 0.034975\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[439]\ttraining's l1: 0.013914\tvalid_1's l1: 0.0349624\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[440]\ttraining's l1: 0.0139011\tvalid_1's l1: 0.0349927\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[441]\ttraining's l1: 0.0138853\tvalid_1's l1: 0.035003\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[442]\ttraining's l1: 0.0138644\tvalid_1's l1: 0.0349934\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[443]\ttraining's l1: 0.0138496\tvalid_1's l1: 0.034965\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[444]\ttraining's l1: 0.0138261\tvalid_1's l1: 0.0349512\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[445]\ttraining's l1: 0.0138033\tvalid_1's l1: 0.0349451\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[446]\ttraining's l1: 0.0137914\tvalid_1's l1: 0.0349672\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[447]\ttraining's l1: 0.0137704\tvalid_1's l1: 0.0349688\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[448]\ttraining's l1: 0.0137601\tvalid_1's l1: 0.0349542\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[449]\ttraining's l1: 0.0137506\tvalid_1's l1: 0.0349454\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[450]\ttraining's l1: 0.0137336\tvalid_1's l1: 0.0349476\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[451]\ttraining's l1: 0.0137078\tvalid_1's l1: 0.0349093\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[452]\ttraining's l1: 0.0136858\tvalid_1's l1: 0.0349145\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[453]\ttraining's l1: 0.0136588\tvalid_1's l1: 0.0349053\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[454]\ttraining's l1: 0.0136365\tvalid_1's l1: 0.0348982\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[455]\ttraining's l1: 0.0136131\tvalid_1's l1: 0.0349001\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[456]\ttraining's l1: 0.0135917\tvalid_1's l1: 0.0348672\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[457]\ttraining's l1: 0.0135783\tvalid_1's l1: 0.0348416\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[458]\ttraining's l1: 0.0135585\tvalid_1's l1: 0.0348479\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[459]\ttraining's l1: 0.0135341\tvalid_1's l1: 0.0348374\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[460]\ttraining's l1: 0.0135116\tvalid_1's l1: 0.0347633\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[461]\ttraining's l1: 0.013494\tvalid_1's l1: 0.034735\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[462]\ttraining's l1: 0.0134782\tvalid_1's l1: 0.0347099\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[463]\ttraining's l1: 0.0134569\tvalid_1's l1: 0.0346972\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[464]\ttraining's l1: 0.0134303\tvalid_1's l1: 0.0346984\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[465]\ttraining's l1: 0.0134174\tvalid_1's l1: 0.0347307\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[466]\ttraining's l1: 0.0133996\tvalid_1's l1: 0.0347041\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[467]\ttraining's l1: 0.0133817\tvalid_1's l1: 0.0347029\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[468]\ttraining's l1: 0.0133527\tvalid_1's l1: 0.0346893\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[469]\ttraining's l1: 0.0133352\tvalid_1's l1: 0.0346628\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[470]\ttraining's l1: 0.0133247\tvalid_1's l1: 0.0346605\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[471]\ttraining's l1: 0.0133176\tvalid_1's l1: 0.0346258\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[472]\ttraining's l1: 0.0132769\tvalid_1's l1: 0.0346398\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[473]\ttraining's l1: 0.0132668\tvalid_1's l1: 0.0346241\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[474]\ttraining's l1: 0.0132427\tvalid_1's l1: 0.0345831\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[475]\ttraining's l1: 0.0132299\tvalid_1's l1: 0.0345522\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[476]\ttraining's l1: 0.0132058\tvalid_1's l1: 0.0345533\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[477]\ttraining's l1: 0.0131975\tvalid_1's l1: 0.0345572\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[478]\ttraining's l1: 0.0131766\tvalid_1's l1: 0.0345488\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[479]\ttraining's l1: 0.0131539\tvalid_1's l1: 0.034543\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[480]\ttraining's l1: 0.0131393\tvalid_1's l1: 0.0345176\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[481]\ttraining's l1: 0.0131317\tvalid_1's l1: 0.034484\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[482]\ttraining's l1: 0.0131156\tvalid_1's l1: 0.03448\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[483]\ttraining's l1: 0.013107\tvalid_1's l1: 0.0345081\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[484]\ttraining's l1: 0.0130958\tvalid_1's l1: 0.0344769\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[485]\ttraining's l1: 0.0130921\tvalid_1's l1: 0.0344702\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[486]\ttraining's l1: 0.0130693\tvalid_1's l1: 0.0344414\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[487]\ttraining's l1: 0.0130498\tvalid_1's l1: 0.0344471\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[488]\ttraining's l1: 0.0130345\tvalid_1's l1: 0.0344253\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[489]\ttraining's l1: 0.0130196\tvalid_1's l1: 0.0344512\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[490]\ttraining's l1: 0.0129959\tvalid_1's l1: 0.0344598\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[491]\ttraining's l1: 0.0129756\tvalid_1's l1: 0.0344435\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[492]\ttraining's l1: 0.0129675\tvalid_1's l1: 0.034471\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[493]\ttraining's l1: 0.0129309\tvalid_1's l1: 0.0344521\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[494]\ttraining's l1: 0.0129108\tvalid_1's l1: 0.0345376\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[495]\ttraining's l1: 0.0128937\tvalid_1's l1: 0.0345327\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[496]\ttraining's l1: 0.0128758\tvalid_1's l1: 0.0345145\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[497]\ttraining's l1: 0.0128526\tvalid_1's l1: 0.0345222\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[498]\ttraining's l1: 0.012841\tvalid_1's l1: 0.0345325\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[499]\ttraining's l1: 0.0128219\tvalid_1's l1: 0.0345371\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[500]\ttraining's l1: 0.0128126\tvalid_1's l1: 0.0345189\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[501]\ttraining's l1: 0.0127959\tvalid_1's l1: 0.0346022\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[502]\ttraining's l1: 0.0127752\tvalid_1's l1: 0.0345878\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[503]\ttraining's l1: 0.012759\tvalid_1's l1: 0.0345649\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[504]\ttraining's l1: 0.0127305\tvalid_1's l1: 0.0345727\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[505]\ttraining's l1: 0.0127094\tvalid_1's l1: 0.0345798\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[506]\ttraining's l1: 0.012697\tvalid_1's l1: 0.0345908\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[507]\ttraining's l1: 0.0126701\tvalid_1's l1: 0.034601\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[508]\ttraining's l1: 0.0126559\tvalid_1's l1: 0.0345956\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[509]\ttraining's l1: 0.0126375\tvalid_1's l1: 0.0345783\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[510]\ttraining's l1: 0.0126285\tvalid_1's l1: 0.0345683\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[511]\ttraining's l1: 0.0126067\tvalid_1's l1: 0.0345668\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[512]\ttraining's l1: 0.0125846\tvalid_1's l1: 0.0345526\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[513]\ttraining's l1: 0.0125749\tvalid_1's l1: 0.0345479\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[514]\ttraining's l1: 0.0125579\tvalid_1's l1: 0.0345525\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[515]\ttraining's l1: 0.0125443\tvalid_1's l1: 0.0345311\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[516]\ttraining's l1: 0.0125378\tvalid_1's l1: 0.034534\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[517]\ttraining's l1: 0.0125065\tvalid_1's l1: 0.0345386\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[518]\ttraining's l1: 0.0124835\tvalid_1's l1: 0.0345459\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[519]\ttraining's l1: 0.0124524\tvalid_1's l1: 0.0345124\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[520]\ttraining's l1: 0.0124317\tvalid_1's l1: 0.0345316\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[521]\ttraining's l1: 0.0124182\tvalid_1's l1: 0.0345268\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[522]\ttraining's l1: 0.0123974\tvalid_1's l1: 0.0345127\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[523]\ttraining's l1: 0.0123824\tvalid_1's l1: 0.0344915\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[524]\ttraining's l1: 0.0123733\tvalid_1's l1: 0.0344897\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[525]\ttraining's l1: 0.0123509\tvalid_1's l1: 0.0344901\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[526]\ttraining's l1: 0.012339\tvalid_1's l1: 0.0345171\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[527]\ttraining's l1: 0.0123197\tvalid_1's l1: 0.0345031\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[528]\ttraining's l1: 0.0123012\tvalid_1's l1: 0.0344989\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[529]\ttraining's l1: 0.0122712\tvalid_1's l1: 0.0344663\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[530]\ttraining's l1: 0.012256\tvalid_1's l1: 0.0344586\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[531]\ttraining's l1: 0.0122402\tvalid_1's l1: 0.0344177\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[532]\ttraining's l1: 0.0122345\tvalid_1's l1: 0.0344311\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[533]\ttraining's l1: 0.0122149\tvalid_1's l1: 0.0344271\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[534]\ttraining's l1: 0.0122035\tvalid_1's l1: 0.0344229\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[535]\ttraining's l1: 0.0121913\tvalid_1's l1: 0.034423\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[536]\ttraining's l1: 0.0121671\tvalid_1's l1: 0.0344038\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[537]\ttraining's l1: 0.0121592\tvalid_1's l1: 0.0344067\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[538]\ttraining's l1: 0.0121303\tvalid_1's l1: 0.0343793\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[539]\ttraining's l1: 0.0121184\tvalid_1's l1: 0.0343922\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[540]\ttraining's l1: 0.0121013\tvalid_1's l1: 0.0343785\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[541]\ttraining's l1: 0.0120949\tvalid_1's l1: 0.0343599\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[542]\ttraining's l1: 0.0120807\tvalid_1's l1: 0.0343558\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[543]\ttraining's l1: 0.0120657\tvalid_1's l1: 0.0343445\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[544]\ttraining's l1: 0.0120485\tvalid_1's l1: 0.0343404\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[545]\ttraining's l1: 0.012037\tvalid_1's l1: 0.0343529\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[546]\ttraining's l1: 0.0120177\tvalid_1's l1: 0.0343609\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[547]\ttraining's l1: 0.0120057\tvalid_1's l1: 0.034354\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[548]\ttraining's l1: 0.0119893\tvalid_1's l1: 0.0343141\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[549]\ttraining's l1: 0.0119774\tvalid_1's l1: 0.0343207\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[550]\ttraining's l1: 0.011957\tvalid_1's l1: 0.0342689\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[551]\ttraining's l1: 0.0119489\tvalid_1's l1: 0.0342674\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[552]\ttraining's l1: 0.0119415\tvalid_1's l1: 0.0342326\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[553]\ttraining's l1: 0.0119249\tvalid_1's l1: 0.0342389\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[554]\ttraining's l1: 0.0118997\tvalid_1's l1: 0.0342193\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[555]\ttraining's l1: 0.0118903\tvalid_1's l1: 0.0342464\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[556]\ttraining's l1: 0.0118747\tvalid_1's l1: 0.0342077\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[557]\ttraining's l1: 0.0118673\tvalid_1's l1: 0.034228\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[558]\ttraining's l1: 0.0118495\tvalid_1's l1: 0.0342281\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[559]\ttraining's l1: 0.011839\tvalid_1's l1: 0.0342167\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[560]\ttraining's l1: 0.0118235\tvalid_1's l1: 0.0341979\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[561]\ttraining's l1: 0.011816\tvalid_1's l1: 0.0341857\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[562]\ttraining's l1: 0.0118011\tvalid_1's l1: 0.034194\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[563]\ttraining's l1: 0.0117937\tvalid_1's l1: 0.034182\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[564]\ttraining's l1: 0.0117864\tvalid_1's l1: 0.0342019\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[565]\ttraining's l1: 0.0117731\tvalid_1's l1: 0.0341902\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[566]\ttraining's l1: 0.0117591\tvalid_1's l1: 0.0341607\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[567]\ttraining's l1: 0.0117491\tvalid_1's l1: 0.0341875\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[568]\ttraining's l1: 0.0117207\tvalid_1's l1: 0.0341614\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[569]\ttraining's l1: 0.0117122\tvalid_1's l1: 0.0341606\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[570]\ttraining's l1: 0.0116958\tvalid_1's l1: 0.0341415\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[571]\ttraining's l1: 0.0116898\tvalid_1's l1: 0.0341343\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[572]\ttraining's l1: 0.011675\tvalid_1's l1: 0.0341419\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[573]\ttraining's l1: 0.0116665\tvalid_1's l1: 0.03414\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[574]\ttraining's l1: 0.0116411\tvalid_1's l1: 0.0341002\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[575]\ttraining's l1: 0.011629\tvalid_1's l1: 0.0340851\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[576]\ttraining's l1: 0.011609\tvalid_1's l1: 0.0340687\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[577]\ttraining's l1: 0.0115964\tvalid_1's l1: 0.0340452\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[578]\ttraining's l1: 0.0115826\tvalid_1's l1: 0.0339763\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[579]\ttraining's l1: 0.0115762\tvalid_1's l1: 0.0339696\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[580]\ttraining's l1: 0.011566\tvalid_1's l1: 0.0339729\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[581]\ttraining's l1: 0.0115552\tvalid_1's l1: 0.0339861\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[582]\ttraining's l1: 0.0115389\tvalid_1's l1: 0.0339444\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[583]\ttraining's l1: 0.0115167\tvalid_1's l1: 0.0339165\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[584]\ttraining's l1: 0.0115099\tvalid_1's l1: 0.0339139\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[585]\ttraining's l1: 0.0114972\tvalid_1's l1: 0.0339034\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[586]\ttraining's l1: 0.0114883\tvalid_1's l1: 0.0339025\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[587]\ttraining's l1: 0.0114794\tvalid_1's l1: 0.0338984\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[588]\ttraining's l1: 0.0114667\tvalid_1's l1: 0.0338881\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[589]\ttraining's l1: 0.0114502\tvalid_1's l1: 0.0338951\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[590]\ttraining's l1: 0.0114372\tvalid_1's l1: 0.033898\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[591]\ttraining's l1: 0.0113996\tvalid_1's l1: 0.0338922\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[592]\ttraining's l1: 0.0113833\tvalid_1's l1: 0.0338854\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[593]\ttraining's l1: 0.0113582\tvalid_1's l1: 0.0338628\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[594]\ttraining's l1: 0.0113541\tvalid_1's l1: 0.0338767\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[595]\ttraining's l1: 0.0113437\tvalid_1's l1: 0.033874\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[596]\ttraining's l1: 0.0113314\tvalid_1's l1: 0.0338461\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[597]\ttraining's l1: 0.0113207\tvalid_1's l1: 0.0338359\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[598]\ttraining's l1: 0.0113054\tvalid_1's l1: 0.0338317\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[599]\ttraining's l1: 0.0112925\tvalid_1's l1: 0.0338242\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[600]\ttraining's l1: 0.0112689\tvalid_1's l1: 0.0338063\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[601]\ttraining's l1: 0.0112562\tvalid_1's l1: 0.0338362\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[602]\ttraining's l1: 0.0112519\tvalid_1's l1: 0.0338046\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[603]\ttraining's l1: 0.0112468\tvalid_1's l1: 0.0338178\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[604]\ttraining's l1: 0.0112169\tvalid_1's l1: 0.033848\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[605]\ttraining's l1: 0.0112127\tvalid_1's l1: 0.0338201\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[606]\ttraining's l1: 0.0111922\tvalid_1's l1: 0.0337993\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[607]\ttraining's l1: 0.0111825\tvalid_1's l1: 0.033819\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[608]\ttraining's l1: 0.0111698\tvalid_1's l1: 0.0338415\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[609]\ttraining's l1: 0.0111587\tvalid_1's l1: 0.0338252\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[610]\ttraining's l1: 0.0111459\tvalid_1's l1: 0.0337904\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[611]\ttraining's l1: 0.0111389\tvalid_1's l1: 0.0337881\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[612]\ttraining's l1: 0.0111344\tvalid_1's l1: 0.0337576\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[613]\ttraining's l1: 0.0111299\tvalid_1's l1: 0.0337705\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[614]\ttraining's l1: 0.0111048\tvalid_1's l1: 0.0337703\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[615]\ttraining's l1: 0.0110955\tvalid_1's l1: 0.0337558\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[616]\ttraining's l1: 0.011083\tvalid_1's l1: 0.0337776\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[617]\ttraining's l1: 0.0110612\tvalid_1's l1: 0.0337701\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[618]\ttraining's l1: 0.0110482\tvalid_1's l1: 0.0337785\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[619]\ttraining's l1: 0.011023\tvalid_1's l1: 0.0337821\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[620]\ttraining's l1: 0.0110115\tvalid_1's l1: 0.0337713\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[621]\ttraining's l1: 0.0109888\tvalid_1's l1: 0.033764\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[622]\ttraining's l1: 0.0109768\tvalid_1's l1: 0.0337528\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[623]\ttraining's l1: 0.0109485\tvalid_1's l1: 0.0337919\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[624]\ttraining's l1: 0.0109359\tvalid_1's l1: 0.0338135\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[625]\ttraining's l1: 0.0109261\tvalid_1's l1: 0.0338218\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[626]\ttraining's l1: 0.0109179\tvalid_1's l1: 0.033792\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[627]\ttraining's l1: 0.010902\tvalid_1's l1: 0.0337821\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[628]\ttraining's l1: 0.0108918\tvalid_1's l1: 0.0337565\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[629]\ttraining's l1: 0.0108848\tvalid_1's l1: 0.0337778\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[630]\ttraining's l1: 0.0108594\tvalid_1's l1: 0.0337916\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[631]\ttraining's l1: 0.0108435\tvalid_1's l1: 0.033765\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[632]\ttraining's l1: 0.0108349\tvalid_1's l1: 0.0337892\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[633]\ttraining's l1: 0.0108258\tvalid_1's l1: 0.033776\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[634]\ttraining's l1: 0.0108109\tvalid_1's l1: 0.033775\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[635]\ttraining's l1: 0.0107911\tvalid_1's l1: 0.0337633\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[636]\ttraining's l1: 0.0107764\tvalid_1's l1: 0.0337703\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[637]\ttraining's l1: 0.0107724\tvalid_1's l1: 0.0337877\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[638]\ttraining's l1: 0.010765\tvalid_1's l1: 0.0337699\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[639]\ttraining's l1: 0.010757\tvalid_1's l1: 0.0337936\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[640]\ttraining's l1: 0.0107475\tvalid_1's l1: 0.0337876\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[641]\ttraining's l1: 0.0107326\tvalid_1's l1: 0.0337677\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[642]\ttraining's l1: 0.0107201\tvalid_1's l1: 0.0337631\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[643]\ttraining's l1: 0.0107077\tvalid_1's l1: 0.0337732\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[644]\ttraining's l1: 0.0106963\tvalid_1's l1: 0.0337777\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[645]\ttraining's l1: 0.0106814\tvalid_1's l1: 0.033781\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[646]\ttraining's l1: 0.0106723\tvalid_1's l1: 0.0337555\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[647]\ttraining's l1: 0.0106613\tvalid_1's l1: 0.033754\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[648]\ttraining's l1: 0.0106489\tvalid_1's l1: 0.0337387\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[649]\ttraining's l1: 0.0106368\tvalid_1's l1: 0.0337218\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[650]\ttraining's l1: 0.0106206\tvalid_1's l1: 0.0337143\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[651]\ttraining's l1: 0.0106096\tvalid_1's l1: 0.0337071\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[652]\ttraining's l1: 0.0105974\tvalid_1's l1: 0.0337055\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[653]\ttraining's l1: 0.0105863\tvalid_1's l1: 0.0337044\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[654]\ttraining's l1: 0.0105751\tvalid_1's l1: 0.0336801\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[655]\ttraining's l1: 0.01055\tvalid_1's l1: 0.0337061\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[656]\ttraining's l1: 0.0105194\tvalid_1's l1: 0.0336981\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[657]\ttraining's l1: 0.0105125\tvalid_1's l1: 0.0336864\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[658]\ttraining's l1: 0.0105088\tvalid_1's l1: 0.0337251\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[659]\ttraining's l1: 0.0105004\tvalid_1's l1: 0.0337067\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[660]\ttraining's l1: 0.0104824\tvalid_1's l1: 0.0336899\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[661]\ttraining's l1: 0.0104816\tvalid_1's l1: 0.0337242\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[662]\ttraining's l1: 0.0104662\tvalid_1's l1: 0.0337169\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[663]\ttraining's l1: 0.0104546\tvalid_1's l1: 0.0337347\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[664]\ttraining's l1: 0.0104454\tvalid_1's l1: 0.0337107\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[665]\ttraining's l1: 0.0104334\tvalid_1's l1: 0.0337118\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[666]\ttraining's l1: 0.0104235\tvalid_1's l1: 0.0337433\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[667]\ttraining's l1: 0.010408\tvalid_1's l1: 0.0337229\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[668]\ttraining's l1: 0.0103999\tvalid_1's l1: 0.0337252\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[669]\ttraining's l1: 0.0103874\tvalid_1's l1: 0.0337062\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[670]\ttraining's l1: 0.0103835\tvalid_1's l1: 0.033744\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[671]\ttraining's l1: 0.0103609\tvalid_1's l1: 0.0337355\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[672]\ttraining's l1: 0.0103397\tvalid_1's l1: 0.0337593\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[673]\ttraining's l1: 0.0103283\tvalid_1's l1: 0.0337478\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[674]\ttraining's l1: 0.0103249\tvalid_1's l1: 0.033785\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[675]\ttraining's l1: 0.0103043\tvalid_1's l1: 0.0337728\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[676]\ttraining's l1: 0.0102935\tvalid_1's l1: 0.0337567\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[677]\ttraining's l1: 0.0102692\tvalid_1's l1: 0.0337905\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[678]\ttraining's l1: 0.0102569\tvalid_1's l1: 0.0337669\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[679]\ttraining's l1: 0.0102445\tvalid_1's l1: 0.0337683\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[680]\ttraining's l1: 0.010233\tvalid_1's l1: 0.0337904\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[681]\ttraining's l1: 0.0102256\tvalid_1's l1: 0.033771\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[682]\ttraining's l1: 0.0102056\tvalid_1's l1: 0.033762\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[683]\ttraining's l1: 0.0102014\tvalid_1's l1: 0.0337655\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[684]\ttraining's l1: 0.010193\tvalid_1's l1: 0.0337566\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[685]\ttraining's l1: 0.0101847\tvalid_1's l1: 0.033733\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[686]\ttraining's l1: 0.0101741\tvalid_1's l1: 0.0337445\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[687]\ttraining's l1: 0.010165\tvalid_1's l1: 0.0337668\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[688]\ttraining's l1: 0.0101521\tvalid_1's l1: 0.0337442\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[689]\ttraining's l1: 0.0101335\tvalid_1's l1: 0.0337324\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[690]\ttraining's l1: 0.0101315\tvalid_1's l1: 0.0337245\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[691]\ttraining's l1: 0.0101164\tvalid_1's l1: 0.0336862\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[692]\ttraining's l1: 0.010106\tvalid_1's l1: 0.0336877\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[693]\ttraining's l1: 0.0100976\tvalid_1's l1: 0.0336647\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[694]\ttraining's l1: 0.0100887\tvalid_1's l1: 0.0336871\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[695]\ttraining's l1: 0.0100819\tvalid_1's l1: 0.0336631\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[696]\ttraining's l1: 0.0100784\tvalid_1's l1: 0.0336994\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[697]\ttraining's l1: 0.0100684\tvalid_1's l1: 0.0337273\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[698]\ttraining's l1: 0.0100565\tvalid_1's l1: 0.0337038\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[699]\ttraining's l1: 0.0100379\tvalid_1's l1: 0.0336926\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[700]\ttraining's l1: 0.0100243\tvalid_1's l1: 0.0337045\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[701]\ttraining's l1: 0.0100136\tvalid_1's l1: 0.0336892\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[702]\ttraining's l1: 0.0100087\tvalid_1's l1: 0.0337055\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[703]\ttraining's l1: 0.00998082\tvalid_1's l1: 0.0336943\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[704]\ttraining's l1: 0.00997006\tvalid_1's l1: 0.0336879\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[705]\ttraining's l1: 0.0099612\tvalid_1's l1: 0.0337173\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[706]\ttraining's l1: 0.00994878\tvalid_1's l1: 0.0337246\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[707]\ttraining's l1: 0.00993472\tvalid_1's l1: 0.0337028\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[708]\ttraining's l1: 0.0099195\tvalid_1's l1: 0.0336993\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[709]\ttraining's l1: 0.00990043\tvalid_1's l1: 0.0336995\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[710]\ttraining's l1: 0.00989124\tvalid_1's l1: 0.0337008\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[711]\ttraining's l1: 0.00987429\tvalid_1's l1: 0.0336809\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[712]\ttraining's l1: 0.00986152\tvalid_1's l1: 0.0336575\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[713]\ttraining's l1: 0.00985213\tvalid_1's l1: 0.03364\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[714]\ttraining's l1: 0.00983371\tvalid_1's l1: 0.033634\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[715]\ttraining's l1: 0.00982491\tvalid_1's l1: 0.0336356\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[716]\ttraining's l1: 0.00981134\tvalid_1's l1: 0.033629\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[717]\ttraining's l1: 0.00980257\tvalid_1's l1: 0.0336147\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[718]\ttraining's l1: 0.00979355\tvalid_1's l1: 0.033604\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[719]\ttraining's l1: 0.00979035\tvalid_1's l1: 0.0336392\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[720]\ttraining's l1: 0.00977812\tvalid_1's l1: 0.0336189\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[721]\ttraining's l1: 0.0097487\tvalid_1's l1: 0.0336251\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[722]\ttraining's l1: 0.00974105\tvalid_1's l1: 0.0336554\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[723]\ttraining's l1: 0.00972833\tvalid_1's l1: 0.03363\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[724]\ttraining's l1: 0.00972221\tvalid_1's l1: 0.0336118\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[725]\ttraining's l1: 0.00971903\tvalid_1's l1: 0.0336467\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[726]\ttraining's l1: 0.00971008\tvalid_1's l1: 0.0336024\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[727]\ttraining's l1: 0.00969442\tvalid_1's l1: 0.033583\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[728]\ttraining's l1: 0.00968692\tvalid_1's l1: 0.0336047\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[729]\ttraining's l1: 0.00967348\tvalid_1's l1: 0.0335939\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[730]\ttraining's l1: 0.00966839\tvalid_1's l1: 0.0336297\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[731]\ttraining's l1: 0.00965615\tvalid_1's l1: 0.0336048\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[732]\ttraining's l1: 0.00964762\tvalid_1's l1: 0.0336262\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[733]\ttraining's l1: 0.00963705\tvalid_1's l1: 0.0336465\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[734]\ttraining's l1: 0.00962862\tvalid_1's l1: 0.0336322\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[735]\ttraining's l1: 0.00962397\tvalid_1's l1: 0.0336166\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[736]\ttraining's l1: 0.00959724\tvalid_1's l1: 0.0336033\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[737]\ttraining's l1: 0.00959024\tvalid_1's l1: 0.0336243\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[738]\ttraining's l1: 0.00958005\tvalid_1's l1: 0.0335999\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[739]\ttraining's l1: 0.00957127\tvalid_1's l1: 0.033609\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[740]\ttraining's l1: 0.00955967\tvalid_1's l1: 0.0335876\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[741]\ttraining's l1: 0.00955574\tvalid_1's l1: 0.0336031\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[742]\ttraining's l1: 0.00954591\tvalid_1's l1: 0.0335895\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[743]\ttraining's l1: 0.00953555\tvalid_1's l1: 0.033565\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[744]\ttraining's l1: 0.00952675\tvalid_1's l1: 0.0335559\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[745]\ttraining's l1: 0.00951949\tvalid_1's l1: 0.0335427\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[746]\ttraining's l1: 0.00951233\tvalid_1's l1: 0.0335682\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[747]\ttraining's l1: 0.00950822\tvalid_1's l1: 0.0335481\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[748]\ttraining's l1: 0.0094991\tvalid_1's l1: 0.0335477\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[749]\ttraining's l1: 0.00948854\tvalid_1's l1: 0.0335773\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[750]\ttraining's l1: 0.00947629\tvalid_1's l1: 0.0335469\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[751]\ttraining's l1: 0.00947295\tvalid_1's l1: 0.0335433\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[752]\ttraining's l1: 0.00946254\tvalid_1's l1: 0.0335655\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[753]\ttraining's l1: 0.0094485\tvalid_1's l1: 0.0335732\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[754]\ttraining's l1: 0.00943836\tvalid_1's l1: 0.0335659\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[755]\ttraining's l1: 0.00942136\tvalid_1's l1: 0.0335489\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[756]\ttraining's l1: 0.00939951\tvalid_1's l1: 0.0335523\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[757]\ttraining's l1: 0.0093907\tvalid_1's l1: 0.033552\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[758]\ttraining's l1: 0.00938291\tvalid_1's l1: 0.0335752\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[759]\ttraining's l1: 0.00937567\tvalid_1's l1: 0.033598\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[760]\ttraining's l1: 0.00936302\tvalid_1's l1: 0.033582\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[761]\ttraining's l1: 0.00935658\tvalid_1's l1: 0.0336152\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[762]\ttraining's l1: 0.00932563\tvalid_1's l1: 0.033628\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[763]\ttraining's l1: 0.00932162\tvalid_1's l1: 0.0336325\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[764]\ttraining's l1: 0.00930769\tvalid_1's l1: 0.033605\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[765]\ttraining's l1: 0.00929316\tvalid_1's l1: 0.033606\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[766]\ttraining's l1: 0.00928377\tvalid_1's l1: 0.0336063\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[767]\ttraining's l1: 0.00927444\tvalid_1's l1: 0.0335833\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[768]\ttraining's l1: 0.00926226\tvalid_1's l1: 0.0335649\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[769]\ttraining's l1: 0.00925444\tvalid_1's l1: 0.033566\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[770]\ttraining's l1: 0.00924821\tvalid_1's l1: 0.0335531\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[771]\ttraining's l1: 0.00923396\tvalid_1's l1: 0.0335446\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[772]\ttraining's l1: 0.00921718\tvalid_1's l1: 0.0335382\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[773]\ttraining's l1: 0.00919534\tvalid_1's l1: 0.0335145\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[774]\ttraining's l1: 0.00918612\tvalid_1's l1: 0.0335148\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[775]\ttraining's l1: 0.00917426\tvalid_1's l1: 0.0334953\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[776]\ttraining's l1: 0.00917079\tvalid_1's l1: 0.0334954\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[777]\ttraining's l1: 0.00916097\tvalid_1's l1: 0.0334883\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[778]\ttraining's l1: 0.0091468\tvalid_1's l1: 0.0334728\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[779]\ttraining's l1: 0.00913755\tvalid_1's l1: 0.0334757\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[780]\ttraining's l1: 0.00912745\tvalid_1's l1: 0.0334667\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[781]\ttraining's l1: 0.00911015\tvalid_1's l1: 0.0334632\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[782]\ttraining's l1: 0.00910116\tvalid_1's l1: 0.0334612\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[783]\ttraining's l1: 0.00909139\tvalid_1's l1: 0.0334572\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[784]\ttraining's l1: 0.00907665\tvalid_1's l1: 0.0334347\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[785]\ttraining's l1: 0.00906588\tvalid_1's l1: 0.0334189\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[786]\ttraining's l1: 0.00905969\tvalid_1's l1: 0.0334393\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[787]\ttraining's l1: 0.009051\tvalid_1's l1: 0.0334299\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[788]\ttraining's l1: 0.00904355\tvalid_1's l1: 0.0334109\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[789]\ttraining's l1: 0.00903957\tvalid_1's l1: 0.0334322\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[790]\ttraining's l1: 0.00902957\tvalid_1's l1: 0.0334165\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[791]\ttraining's l1: 0.0090227\tvalid_1's l1: 0.0334376\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[792]\ttraining's l1: 0.00901328\tvalid_1's l1: 0.0334308\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[793]\ttraining's l1: 0.00900098\tvalid_1's l1: 0.0334229\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[794]\ttraining's l1: 0.00899319\tvalid_1's l1: 0.0334074\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[795]\ttraining's l1: 0.00898232\tvalid_1's l1: 0.0334075\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[796]\ttraining's l1: 0.00897534\tvalid_1's l1: 0.0334076\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[797]\ttraining's l1: 0.00896255\tvalid_1's l1: 0.0334171\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[798]\ttraining's l1: 0.00894559\tvalid_1's l1: 0.0333963\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[799]\ttraining's l1: 0.00894245\tvalid_1's l1: 0.0334312\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[800]\ttraining's l1: 0.00893644\tvalid_1's l1: 0.0334323\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[801]\ttraining's l1: 0.00892825\tvalid_1's l1: 0.0334112\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[802]\ttraining's l1: 0.00891511\tvalid_1's l1: 0.0333956\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[803]\ttraining's l1: 0.00891278\tvalid_1's l1: 0.0334115\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[804]\ttraining's l1: 0.00890117\tvalid_1's l1: 0.0333952\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[805]\ttraining's l1: 0.00889453\tvalid_1's l1: 0.033395\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[806]\ttraining's l1: 0.00888522\tvalid_1's l1: 0.0334019\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[807]\ttraining's l1: 0.00887542\tvalid_1's l1: 0.0334048\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[808]\ttraining's l1: 0.00886727\tvalid_1's l1: 0.0333955\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[809]\ttraining's l1: 0.00886136\tvalid_1's l1: 0.0333987\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[810]\ttraining's l1: 0.00885253\tvalid_1's l1: 0.0333891\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[811]\ttraining's l1: 0.00884695\tvalid_1's l1: 0.0334109\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[812]\ttraining's l1: 0.00883677\tvalid_1's l1: 0.0333891\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[813]\ttraining's l1: 0.00882986\tvalid_1's l1: 0.0333698\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[814]\ttraining's l1: 0.00881862\tvalid_1's l1: 0.0333623\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[815]\ttraining's l1: 0.00880099\tvalid_1's l1: 0.0333797\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[816]\ttraining's l1: 0.0087895\tvalid_1's l1: 0.0333805\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[817]\ttraining's l1: 0.00877981\tvalid_1's l1: 0.0333791\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[818]\ttraining's l1: 0.00876559\tvalid_1's l1: 0.0333649\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[819]\ttraining's l1: 0.00875683\tvalid_1's l1: 0.0333697\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[820]\ttraining's l1: 0.00874872\tvalid_1's l1: 0.0333853\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[821]\ttraining's l1: 0.00873146\tvalid_1's l1: 0.0333718\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[822]\ttraining's l1: 0.00871068\tvalid_1's l1: 0.0333674\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[823]\ttraining's l1: 0.00869328\tvalid_1's l1: 0.033353\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[824]\ttraining's l1: 0.008686\tvalid_1's l1: 0.0333528\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[825]\ttraining's l1: 0.00868024\tvalid_1's l1: 0.0333347\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[826]\ttraining's l1: 0.00867263\tvalid_1's l1: 0.033338\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[827]\ttraining's l1: 0.00866488\tvalid_1's l1: 0.0333199\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[828]\ttraining's l1: 0.00865679\tvalid_1's l1: 0.0333247\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[829]\ttraining's l1: 0.00865087\tvalid_1's l1: 0.033349\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[830]\ttraining's l1: 0.00864493\tvalid_1's l1: 0.0333348\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[831]\ttraining's l1: 0.00863743\tvalid_1's l1: 0.0333381\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[832]\ttraining's l1: 0.00863042\tvalid_1's l1: 0.0333254\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[833]\ttraining's l1: 0.00862055\tvalid_1's l1: 0.0333048\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[834]\ttraining's l1: 0.00861339\tvalid_1's l1: 0.033308\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[835]\ttraining's l1: 0.00860662\tvalid_1's l1: 0.0332956\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[836]\ttraining's l1: 0.00859945\tvalid_1's l1: 0.0332921\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[837]\ttraining's l1: 0.00858109\tvalid_1's l1: 0.0332731\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[838]\ttraining's l1: 0.00856701\tvalid_1's l1: 0.0332576\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[839]\ttraining's l1: 0.00855966\tvalid_1's l1: 0.0332731\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[840]\ttraining's l1: 0.00855297\tvalid_1's l1: 0.0332722\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[841]\ttraining's l1: 0.00853631\tvalid_1's l1: 0.0332844\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[842]\ttraining's l1: 0.00852926\tvalid_1's l1: 0.0332638\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[843]\ttraining's l1: 0.00852548\tvalid_1's l1: 0.0332668\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[844]\ttraining's l1: 0.00850826\tvalid_1's l1: 0.0332665\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[845]\ttraining's l1: 0.00850181\tvalid_1's l1: 0.033268\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[846]\ttraining's l1: 0.00849723\tvalid_1's l1: 0.0332569\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[847]\ttraining's l1: 0.00848774\tvalid_1's l1: 0.0332428\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[848]\ttraining's l1: 0.00848396\tvalid_1's l1: 0.0332457\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[849]\ttraining's l1: 0.00847568\tvalid_1's l1: 0.0332537\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[850]\ttraining's l1: 0.0084684\tvalid_1's l1: 0.0332451\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[851]\ttraining's l1: 0.00845583\tvalid_1's l1: 0.0332632\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[852]\ttraining's l1: 0.00845195\tvalid_1's l1: 0.0332821\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[853]\ttraining's l1: 0.00844431\tvalid_1's l1: 0.0332618\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[854]\ttraining's l1: 0.00843376\tvalid_1's l1: 0.0332562\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[855]\ttraining's l1: 0.00842688\tvalid_1's l1: 0.0332391\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[856]\ttraining's l1: 0.00842197\tvalid_1's l1: 0.0332618\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[857]\ttraining's l1: 0.00841119\tvalid_1's l1: 0.033253\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[858]\ttraining's l1: 0.00840732\tvalid_1's l1: 0.0332557\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[859]\ttraining's l1: 0.00839113\tvalid_1's l1: 0.0332685\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[860]\ttraining's l1: 0.0083827\tvalid_1's l1: 0.033255\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[861]\ttraining's l1: 0.00837578\tvalid_1's l1: 0.0332406\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[862]\ttraining's l1: 0.00835868\tvalid_1's l1: 0.0332521\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[863]\ttraining's l1: 0.00834269\tvalid_1's l1: 0.0332548\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[864]\ttraining's l1: 0.0083334\tvalid_1's l1: 0.0332369\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[865]\ttraining's l1: 0.00832426\tvalid_1's l1: 0.0332242\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[866]\ttraining's l1: 0.00831367\tvalid_1's l1: 0.0332302\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[867]\ttraining's l1: 0.00829399\tvalid_1's l1: 0.0332263\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[868]\ttraining's l1: 0.00828484\tvalid_1's l1: 0.0332289\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[869]\ttraining's l1: 0.00827795\tvalid_1's l1: 0.0332203\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[870]\ttraining's l1: 0.00827597\tvalid_1's l1: 0.0332407\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[871]\ttraining's l1: 0.0082677\tvalid_1's l1: 0.033227\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[872]\ttraining's l1: 0.00825947\tvalid_1's l1: 0.0332417\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[873]\ttraining's l1: 0.00825119\tvalid_1's l1: 0.0332278\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[874]\ttraining's l1: 0.0082435\tvalid_1's l1: 0.0332191\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[875]\ttraining's l1: 0.00822835\tvalid_1's l1: 0.0332065\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[876]\ttraining's l1: 0.00822312\tvalid_1's l1: 0.0332094\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[877]\ttraining's l1: 0.00820974\tvalid_1's l1: 0.0331966\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[878]\ttraining's l1: 0.0082028\tvalid_1's l1: 0.0331893\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[879]\ttraining's l1: 0.00818742\tvalid_1's l1: 0.0331918\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[880]\ttraining's l1: 0.00817342\tvalid_1's l1: 0.0331793\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[881]\ttraining's l1: 0.00816485\tvalid_1's l1: 0.033163\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[882]\ttraining's l1: 0.00815875\tvalid_1's l1: 0.0331498\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[883]\ttraining's l1: 0.00815077\tvalid_1's l1: 0.0331784\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[884]\ttraining's l1: 0.00813986\tvalid_1's l1: 0.0331835\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[885]\ttraining's l1: 0.00813122\tvalid_1's l1: 0.0332114\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[886]\ttraining's l1: 0.00812329\tvalid_1's l1: 0.0332134\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[887]\ttraining's l1: 0.00811593\tvalid_1's l1: 0.033216\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[888]\ttraining's l1: 0.00810868\tvalid_1's l1: 0.0332072\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[889]\ttraining's l1: 0.00810336\tvalid_1's l1: 0.033206\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[890]\ttraining's l1: 0.00809585\tvalid_1's l1: 0.0332208\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[891]\ttraining's l1: 0.00808914\tvalid_1's l1: 0.0332048\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[892]\ttraining's l1: 0.00808141\tvalid_1's l1: 0.033197\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[893]\ttraining's l1: 0.00807404\tvalid_1's l1: 0.0332187\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[894]\ttraining's l1: 0.00806815\tvalid_1's l1: 0.0332388\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[895]\ttraining's l1: 0.00805902\tvalid_1's l1: 0.0332233\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[896]\ttraining's l1: 0.00805111\tvalid_1's l1: 0.0332213\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[897]\ttraining's l1: 0.00804592\tvalid_1's l1: 0.0332201\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[898]\ttraining's l1: 0.00803409\tvalid_1's l1: 0.0332365\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[899]\ttraining's l1: 0.00802259\tvalid_1's l1: 0.0332186\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[900]\ttraining's l1: 0.00801537\tvalid_1's l1: 0.0332402\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[901]\ttraining's l1: 0.00800972\tvalid_1's l1: 0.0332582\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[902]\ttraining's l1: 0.00800333\tvalid_1's l1: 0.0332456\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[903]\ttraining's l1: 0.00798943\tvalid_1's l1: 0.0332578\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[904]\ttraining's l1: 0.0079793\tvalid_1's l1: 0.0332591\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[905]\ttraining's l1: 0.00797116\tvalid_1's l1: 0.0332424\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[906]\ttraining's l1: 0.00796316\tvalid_1's l1: 0.033226\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[907]\ttraining's l1: 0.00795597\tvalid_1's l1: 0.033244\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[908]\ttraining's l1: 0.0079465\tvalid_1's l1: 0.0332254\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[909]\ttraining's l1: 0.00794099\tvalid_1's l1: 0.0332133\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[910]\ttraining's l1: 0.00792834\tvalid_1's l1: 0.0332154\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[911]\ttraining's l1: 0.00792304\tvalid_1's l1: 0.0332365\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[912]\ttraining's l1: 0.00791057\tvalid_1's l1: 0.0332384\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[913]\ttraining's l1: 0.00790349\tvalid_1's l1: 0.0332419\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[914]\ttraining's l1: 0.00789277\tvalid_1's l1: 0.033252\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[915]\ttraining's l1: 0.00789007\tvalid_1's l1: 0.0332894\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[916]\ttraining's l1: 0.00788205\tvalid_1's l1: 0.0332731\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[917]\ttraining's l1: 0.00786532\tvalid_1's l1: 0.0332679\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[918]\ttraining's l1: 0.00785201\tvalid_1's l1: 0.0332784\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[919]\ttraining's l1: 0.00784544\tvalid_1's l1: 0.0332647\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[920]\ttraining's l1: 0.00784032\tvalid_1's l1: 0.0332857\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[921]\ttraining's l1: 0.0078355\tvalid_1's l1: 0.0332777\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[922]\ttraining's l1: 0.00782106\tvalid_1's l1: 0.0332803\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[923]\ttraining's l1: 0.00781722\tvalid_1's l1: 0.0332686\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[924]\ttraining's l1: 0.00780484\tvalid_1's l1: 0.0332733\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[925]\ttraining's l1: 0.00779535\tvalid_1's l1: 0.0332673\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[926]\ttraining's l1: 0.00779343\tvalid_1's l1: 0.0332893\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[927]\ttraining's l1: 0.00778773\tvalid_1's l1: 0.033289\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[928]\ttraining's l1: 0.00778114\tvalid_1's l1: 0.0332736\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[929]\ttraining's l1: 0.00777327\tvalid_1's l1: 0.0332877\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[930]\ttraining's l1: 0.00776207\tvalid_1's l1: 0.0333033\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[931]\ttraining's l1: 0.00775861\tvalid_1's l1: 0.0333224\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[932]\ttraining's l1: 0.00775409\tvalid_1's l1: 0.0333059\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[933]\ttraining's l1: 0.00774052\tvalid_1's l1: 0.0333019\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[934]\ttraining's l1: 0.00773685\tvalid_1's l1: 0.0332851\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[935]\ttraining's l1: 0.00773149\tvalid_1's l1: 0.033285\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[936]\ttraining's l1: 0.00772038\tvalid_1's l1: 0.0333004\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[937]\ttraining's l1: 0.00771848\tvalid_1's l1: 0.0333241\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[938]\ttraining's l1: 0.00771283\tvalid_1's l1: 0.0333079\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[939]\ttraining's l1: 0.00769256\tvalid_1's l1: 0.0333045\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[940]\ttraining's l1: 0.00768103\tvalid_1's l1: 0.0332985\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[941]\ttraining's l1: 0.00766899\tvalid_1's l1: 0.0333073\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[942]\ttraining's l1: 0.00766212\tvalid_1's l1: 0.0333242\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[943]\ttraining's l1: 0.00765774\tvalid_1's l1: 0.0333437\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[944]\ttraining's l1: 0.00764398\tvalid_1's l1: 0.033332\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[945]\ttraining's l1: 0.00764145\tvalid_1's l1: 0.0333514\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[946]\ttraining's l1: 0.00763301\tvalid_1's l1: 0.0333345\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[947]\ttraining's l1: 0.00762492\tvalid_1's l1: 0.033361\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[948]\ttraining's l1: 0.00762091\tvalid_1's l1: 0.0333434\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[949]\ttraining's l1: 0.00761674\tvalid_1's l1: 0.0333607\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[950]\ttraining's l1: 0.00761312\tvalid_1's l1: 0.0333444\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[951]\ttraining's l1: 0.00760785\tvalid_1's l1: 0.0333537\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[952]\ttraining's l1: 0.00760574\tvalid_1's l1: 0.0333898\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[953]\ttraining's l1: 0.00759982\tvalid_1's l1: 0.0333804\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[954]\ttraining's l1: 0.00758697\tvalid_1's l1: 0.0333687\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[955]\ttraining's l1: 0.00758253\tvalid_1's l1: 0.0333542\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[956]\ttraining's l1: 0.00758158\tvalid_1's l1: 0.03339\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[957]\ttraining's l1: 0.00757074\tvalid_1's l1: 0.0333809\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[958]\ttraining's l1: 0.0075705\tvalid_1's l1: 0.0334019\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[959]\ttraining's l1: 0.00755856\tvalid_1's l1: 0.033396\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[960]\ttraining's l1: 0.00754607\tvalid_1's l1: 0.0333975\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[961]\ttraining's l1: 0.00754497\tvalid_1's l1: 0.0334002\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[962]\ttraining's l1: 0.00753231\tvalid_1's l1: 0.0333956\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[963]\ttraining's l1: 0.00752682\tvalid_1's l1: 0.0333914\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[964]\ttraining's l1: 0.00752035\tvalid_1's l1: 0.0333907\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[965]\ttraining's l1: 0.00750944\tvalid_1's l1: 0.0334056\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[966]\ttraining's l1: 0.00750841\tvalid_1's l1: 0.0334217\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[967]\ttraining's l1: 0.00750398\tvalid_1's l1: 0.0334071\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[968]\ttraining's l1: 0.00749763\tvalid_1's l1: 0.0333998\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[969]\ttraining's l1: 0.00749396\tvalid_1's l1: 0.03342\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[970]\ttraining's l1: 0.00748137\tvalid_1's l1: 0.0334178\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[971]\ttraining's l1: 0.00747163\tvalid_1's l1: 0.0334261\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[972]\ttraining's l1: 0.00747057\tvalid_1's l1: 0.0334465\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[973]\ttraining's l1: 0.00745848\tvalid_1's l1: 0.0334353\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[974]\ttraining's l1: 0.00744824\tvalid_1's l1: 0.033452\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[975]\ttraining's l1: 0.0074367\tvalid_1's l1: 0.0334424\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[976]\ttraining's l1: 0.00743123\tvalid_1's l1: 0.0334266\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[977]\ttraining's l1: 0.00742199\tvalid_1's l1: 0.0334129\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[978]\ttraining's l1: 0.00741311\tvalid_1's l1: 0.0334036\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[979]\ttraining's l1: 0.00741287\tvalid_1's l1: 0.0334245\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[980]\ttraining's l1: 0.00740621\tvalid_1's l1: 0.0334154\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[981]\ttraining's l1: 0.00739818\tvalid_1's l1: 0.033425\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[982]\ttraining's l1: 0.00739258\tvalid_1's l1: 0.0334096\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[983]\ttraining's l1: 0.00738675\tvalid_1's l1: 0.0334261\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[984]\ttraining's l1: 0.00737844\tvalid_1's l1: 0.0334168\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[985]\ttraining's l1: 0.00737195\tvalid_1's l1: 0.0334079\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[986]\ttraining's l1: 0.00736445\tvalid_1's l1: 0.0333937\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[987]\ttraining's l1: 0.00735237\tvalid_1's l1: 0.0334045\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[988]\ttraining's l1: 0.00734054\tvalid_1's l1: 0.0334064\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[989]\ttraining's l1: 0.0073391\tvalid_1's l1: 0.0334221\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[990]\ttraining's l1: 0.00733473\tvalid_1's l1: 0.0334162\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[991]\ttraining's l1: 0.00733012\tvalid_1's l1: 0.0334011\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[992]\ttraining's l1: 0.00733005\tvalid_1's l1: 0.0334167\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[993]\ttraining's l1: 0.00731772\tvalid_1's l1: 0.0334167\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[994]\ttraining's l1: 0.00730808\tvalid_1's l1: 0.0333985\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[995]\ttraining's l1: 0.0073015\tvalid_1's l1: 0.0333834\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[996]\ttraining's l1: 0.00729342\tvalid_1's l1: 0.033409\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[997]\ttraining's l1: 0.00729354\tvalid_1's l1: 0.0334028\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[998]\ttraining's l1: 0.00728604\tvalid_1's l1: 0.033391\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[999]\ttraining's l1: 0.00727383\tvalid_1's l1: 0.0333878\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1000]\ttraining's l1: 0.00726702\tvalid_1's l1: 0.0333731\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1001]\ttraining's l1: 0.00725927\tvalid_1's l1: 0.0333736\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1002]\ttraining's l1: 0.00724624\tvalid_1's l1: 0.0333727\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1003]\ttraining's l1: 0.00724044\tvalid_1's l1: 0.0333558\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1004]\ttraining's l1: 0.00723077\tvalid_1's l1: 0.0333378\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1005]\ttraining's l1: 0.00722557\tvalid_1's l1: 0.0333395\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1006]\ttraining's l1: 0.00721362\tvalid_1's l1: 0.0333287\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1007]\ttraining's l1: 0.00721301\tvalid_1's l1: 0.0333225\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1008]\ttraining's l1: 0.00720083\tvalid_1's l1: 0.0333167\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1009]\ttraining's l1: 0.00719415\tvalid_1's l1: 0.033329\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1010]\ttraining's l1: 0.00719485\tvalid_1's l1: 0.0333497\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1011]\ttraining's l1: 0.00718308\tvalid_1's l1: 0.033339\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1012]\ttraining's l1: 0.00717813\tvalid_1's l1: 0.03334\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1013]\ttraining's l1: 0.00716697\tvalid_1's l1: 0.0333453\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1014]\ttraining's l1: 0.00715698\tvalid_1's l1: 0.0333431\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1015]\ttraining's l1: 0.00714843\tvalid_1's l1: 0.0333291\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1016]\ttraining's l1: 0.00714259\tvalid_1's l1: 0.0333323\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1017]\ttraining's l1: 0.00713152\tvalid_1's l1: 0.0333418\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1018]\ttraining's l1: 0.00712286\tvalid_1's l1: 0.0333201\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1019]\ttraining's l1: 0.00711209\tvalid_1's l1: 0.0333227\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1020]\ttraining's l1: 0.00710566\tvalid_1's l1: 0.0333468\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1021]\ttraining's l1: 0.00710247\tvalid_1's l1: 0.0333626\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1022]\ttraining's l1: 0.00708664\tvalid_1's l1: 0.0333632\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1023]\ttraining's l1: 0.00707479\tvalid_1's l1: 0.0333569\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1024]\ttraining's l1: 0.00707315\tvalid_1's l1: 0.0333743\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1025]\ttraining's l1: 0.00706205\tvalid_1's l1: 0.0333573\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1026]\ttraining's l1: 0.00705537\tvalid_1's l1: 0.0333575\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1027]\ttraining's l1: 0.00705357\tvalid_1's l1: 0.0333573\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1028]\ttraining's l1: 0.00705362\tvalid_1's l1: 0.033377\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1029]\ttraining's l1: 0.00704287\tvalid_1's l1: 0.033374\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1030]\ttraining's l1: 0.00703242\tvalid_1's l1: 0.0333743\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1031]\ttraining's l1: 0.00702686\tvalid_1's l1: 0.0333677\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1032]\ttraining's l1: 0.00701672\tvalid_1's l1: 0.0333576\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1033]\ttraining's l1: 0.0070152\tvalid_1's l1: 0.0333907\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1034]\ttraining's l1: 0.00700448\tvalid_1's l1: 0.0333825\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1035]\ttraining's l1: 0.00700018\tvalid_1's l1: 0.0333892\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1036]\ttraining's l1: 0.00698923\tvalid_1's l1: 0.0333836\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1037]\ttraining's l1: 0.00697353\tvalid_1's l1: 0.0333704\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1038]\ttraining's l1: 0.00696811\tvalid_1's l1: 0.0333859\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1039]\ttraining's l1: 0.00696276\tvalid_1's l1: 0.0334093\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1040]\ttraining's l1: 0.00695199\tvalid_1's l1: 0.0334109\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1041]\ttraining's l1: 0.00694374\tvalid_1's l1: 0.0334092\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1042]\ttraining's l1: 0.00693461\tvalid_1's l1: 0.0334022\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1043]\ttraining's l1: 0.00692542\tvalid_1's l1: 0.0334111\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1044]\ttraining's l1: 0.00691841\tvalid_1's l1: 0.0333929\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1045]\ttraining's l1: 0.0069064\tvalid_1's l1: 0.0333773\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1046]\ttraining's l1: 0.00690382\tvalid_1's l1: 0.0333715\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1047]\ttraining's l1: 0.00689848\tvalid_1's l1: 0.0333866\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1048]\ttraining's l1: 0.00688782\tvalid_1's l1: 0.0333785\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1049]\ttraining's l1: 0.00688229\tvalid_1's l1: 0.0333667\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1050]\ttraining's l1: 0.00687366\tvalid_1's l1: 0.0333598\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1051]\ttraining's l1: 0.00686915\tvalid_1's l1: 0.0333668\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1052]\ttraining's l1: 0.0068606\tvalid_1's l1: 0.0333682\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1053]\ttraining's l1: 0.0068556\tvalid_1's l1: 0.033359\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1054]\ttraining's l1: 0.00684089\tvalid_1's l1: 0.0333565\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1055]\ttraining's l1: 0.00683128\tvalid_1's l1: 0.0333647\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1056]\ttraining's l1: 0.0068228\tvalid_1's l1: 0.0333669\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1057]\ttraining's l1: 0.00681898\tvalid_1's l1: 0.0333819\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1058]\ttraining's l1: 0.00681301\tvalid_1's l1: 0.0333732\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1059]\ttraining's l1: 0.00681013\tvalid_1's l1: 0.0333676\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1060]\ttraining's l1: 0.00679814\tvalid_1's l1: 0.0333555\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1061]\ttraining's l1: 0.00679422\tvalid_1's l1: 0.0333704\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1062]\ttraining's l1: 0.00678869\tvalid_1's l1: 0.0333674\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1063]\ttraining's l1: 0.00678247\tvalid_1's l1: 0.0333698\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1064]\ttraining's l1: 0.00677263\tvalid_1's l1: 0.0333696\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1065]\ttraining's l1: 0.00676754\tvalid_1's l1: 0.0333692\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1066]\ttraining's l1: 0.00675808\tvalid_1's l1: 0.0333514\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1067]\ttraining's l1: 0.00674681\tvalid_1's l1: 0.0333416\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1068]\ttraining's l1: 0.00674085\tvalid_1's l1: 0.0333561\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1069]\ttraining's l1: 0.00674123\tvalid_1's l1: 0.0333497\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1070]\ttraining's l1: 0.0067391\tvalid_1's l1: 0.0333704\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1071]\ttraining's l1: 0.00672853\tvalid_1's l1: 0.0333536\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1072]\ttraining's l1: 0.00672473\tvalid_1's l1: 0.0333597\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1073]\ttraining's l1: 0.00671835\tvalid_1's l1: 0.0333731\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1074]\ttraining's l1: 0.00671692\tvalid_1's l1: 0.0333874\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1075]\ttraining's l1: 0.00671183\tvalid_1's l1: 0.033374\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1076]\ttraining's l1: 0.00670213\tvalid_1's l1: 0.0333642\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1077]\ttraining's l1: 0.00669445\tvalid_1's l1: 0.0333588\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1078]\ttraining's l1: 0.00669413\tvalid_1's l1: 0.0333611\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1079]\ttraining's l1: 0.00668452\tvalid_1's l1: 0.0333594\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1080]\ttraining's l1: 0.00667909\tvalid_1's l1: 0.0333587\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1081]\ttraining's l1: 0.00666681\tvalid_1's l1: 0.0333426\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[1082]\ttraining's l1: 0.00666418\tvalid_1's l1: 0.0333606\n",
            "Early stopping, best iteration is:\n",
            "[882]\ttraining's l1: 0.00815875\tvalid_1's l1: 0.0331498\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "3.373344925444753"
            ]
          },
          "metadata": {},
          "execution_count": 35
        }
      ],
      "source": [
        "# LightGBM parameters\n",
        "lgb_params = {'metric': {'mae'},\n",
        "              'num_leaves': 10,\n",
        "              'learning_rate': 0.02,\n",
        "              'feature_fraction': 0.8,\n",
        "              'feature_name' : 189,\n",
        "              'max_depth': 5,\n",
        "              'verbose': 0,\n",
        "              'num_boost_round': 10000, \n",
        "              'early_stopping_rounds': 200, \n",
        "              'nthread': -1}\n",
        "\n",
        "lgbtrain = lgb.Dataset(data=X_train, label=Y_train, feature_name=cols[:189])\n",
        "#lgbtrain = lgb.Dataset(data=X_train, label=Y_train, feature_name=cols)\n",
        "#lgbval = lgb.Dataset(data=X_test, label=Y_test, reference=lgbtrain, feature_name=cols)\n",
        "\n",
        "#model = lgb.train(lgb_params, lgbtrain,\n",
        "          #        valid_sets=[lgbtrain, lgbval],\n",
        "         #         num_boost_round=lgb_params['num_boost_round'],\n",
        "           #       early_stopping_rounds=lgb_params['early_stopping_rounds'],\n",
        "            #      feval=lgbm_smape, \n",
        "             #     verbose_eval=100)\n",
        "\n",
        "lgbval = lgb.Dataset(data=X_test, label=Y_test, reference=lgbtrain)\n",
        "model = lgb.train(lgb_params, lgbtrain,\n",
        "                   valid_sets=[lgbtrain, lgbval],\n",
        "                   num_boost_round=lgb_params['num_boost_round'])\n",
        "\n",
        "y_pred_test = model.predict(X_test, num_iteration=model.best_iteration)\n",
        "\n",
        "# percentage of test error\n",
        "smape(np.expm1(y_pred_test), np.expm1(Y_test))\n"
      ],
      "id": "45df8046"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "31ab59df"
      },
      "source": [
        "### Feature Importance"
      ],
      "id": "31ab59df"
    },
    {
      "cell_type": "code",
      "execution_count": 36,
      "metadata": {
        "id": "7631841c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "b48c8b22-3717-4336-f40e-e240936e7a1a"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "                       feature  split       gain\n",
            "51   sales_ewm_alpha_05_lag_50    160  48.167349\n",
            "46   sales_ewm_alpha_07_lag_50    183  12.087733\n",
            "4                  day_of_year    670   9.736439\n",
            "2                        month    173   4.632800\n",
            "5                 week_of_year    315   3.529493\n",
            "17          sales_roll_mean_61    154   3.206748\n",
            "41   sales_ewm_alpha_08_lag_50     74   2.858602\n",
            "36   sales_ewm_alpha_09_lag_50     32   2.475369\n",
            "14               sales_lag_105    217   2.096715\n",
            "22  sales_ewm_alpha_099_lag_10    108   1.612017\n",
            "25  sales_ewm_alpha_099_lag_40    136   1.263724\n",
            "23  sales_ewm_alpha_099_lag_20    144   1.180727\n",
            "47   sales_ewm_alpha_05_lag_10    126   0.998215\n",
            "18          sales_roll_mean_91     59   0.940485\n",
            "3                 day_of_month    438   0.451319\n",
            "16          sales_roll_mean_31    232   0.374652\n",
            "43   sales_ewm_alpha_07_lag_20    153   0.371154\n",
            "12                sales_lag_91    184   0.363785\n",
            "21         sales_roll_mean_112    125   0.321036\n",
            "24  sales_ewm_alpha_099_lag_30    127   0.295586\n",
            "15               sales_lag_112     51   0.291773\n",
            "10                sales_lag_31    133   0.271112\n",
            "48   sales_ewm_alpha_05_lag_20    109   0.255273\n",
            "27  sales_ewm_alpha_095_lag_10     50   0.233311\n",
            "30  sales_ewm_alpha_095_lag_40    112   0.212002\n",
            "11                sales_lag_61     70   0.198242\n",
            "28  sales_ewm_alpha_095_lag_20     72   0.191002\n",
            "34   sales_ewm_alpha_09_lag_30     34   0.158125\n",
            "29  sales_ewm_alpha_095_lag_30     40   0.155451\n",
            "33   sales_ewm_alpha_09_lag_20     50   0.121011\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1000x1000 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "def plot_lgb_importances(model, plot=False, num=10):\n",
        "\n",
        "    gain = model.feature_importance('gain')\n",
        "    feat_imp = pd.DataFrame({'feature': model.feature_name(),\n",
        "                             'split': model.feature_importance('split'),\n",
        "                             'gain': 100 * gain / gain.sum()}).sort_values('gain', ascending=False)\n",
        "    if plot:\n",
        "        plt.figure(figsize=(10, 10))\n",
        "        sns.set(font_scale=1)\n",
        "        sns.barplot(x=\"gain\", y=\"feature\", data=feat_imp[0:25])\n",
        "        plt.title('feature')\n",
        "        plt.tight_layout()\n",
        "        plt.show()\n",
        "    else:\n",
        "        print(feat_imp.head(num))\n",
        "\n",
        "\n",
        "plot_lgb_importances(model, num=30)\n",
        "#plot_lgb_importances(model, num=30, plot=True)\n",
        "\n",
        "lgb.plot_importance(model, max_num_features=20, figsize=(10, 10), importance_type=\"gain\")\n",
        "plt.show()\n"
      ],
      "id": "7631841c"
    },
    {
      "cell_type": "code",
      "execution_count": 37,
      "metadata": {
        "id": "9b4c5ba4",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "81d7be62-631b-47b8-a5d8-d3a3b2d9a197"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[LightGBM] [Warning] Auto-choosing col-wise multi-threading, the overhead of testing was 0.000144 seconds.\n",
            "You can set `force_col_wise=true` to remove the overhead.\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n",
            "[LightGBM] [Warning] No further splits with positive gain, best gain: -inf\n"
          ]
        }
      ],
      "source": [
        "# Final Model\n",
        "\n",
        "lgb_params = {'metric': {'mae'},\n",
        "              'num_leaves': 10,\n",
        "              'learning_rate': 0.02,\n",
        "              'feature_fraction': 0.8,\n",
        "              'max_depth': 5,\n",
        "              'verbose': 0,\n",
        "              'nthread': -1,\n",
        "              \"num_boost_round\": model.best_iteration}\n",
        "\n",
        "# LightGBM dataset\n",
        "lgbtrain_all = lgb.Dataset(data=X_train, label=Y_train, feature_name=cols)\n",
        "\n",
        "model = lgb.train(lgb_params, lgbtrain_all, num_boost_round=model.best_iteration)\n",
        "test_preds = model.predict(X_test, num_iteration=model.best_iteration)\n"
      ],
      "id": "9b4c5ba4"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "81622a96"
      },
      "source": [
        "## Here is the prediction!"
      ],
      "id": "81622a96"
    },
    {
      "cell_type": "code",
      "execution_count": 38,
      "metadata": {
        "id": "74bb0356",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 462
        },
        "outputId": "8ab28bf5-be7d-434e-ed7c-82f3cb73f122"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 2000x900 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "# 1 year actual and predicted values\n",
        "forecast = pd.DataFrame({\"date\":test[\"date\"],\n",
        "                        \"store\":test[\"store\"],\n",
        "                        \"item\":test[\"item\"],\n",
        "                        \"sales\":test_preds\n",
        "                        })\n",
        "\n",
        "df.set_index(\"date\").sales.plot(figsize = (20,9),legend=True, label = \"Actual\")\n",
        "forecast.set_index(\"date\").sales.plot(legend=True, label = \"Predict\")\n",
        "plt.show()"
      ],
      "id": "74bb0356"
    }
  ],
  "metadata": {
    "colab": {
      "provenance": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.7.10"
    },
    "papermill": {
      "default_parameters": {},
      "duration": 2170.42965,
      "end_time": "2021-10-14T21:06:34.632238",
      "environment_variables": {},
      "exception": null,
      "input_path": "__notebook__.ipynb",
      "output_path": "__notebook__.ipynb",
      "parameters": {},
      "start_time": "2021-10-14T20:30:24.202588",
      "version": "2.3.3"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 5
}