Skip to content

This repository showcases my portfolio of advanced projects and analyses completed during my Master's in Applied Statistics and Data Science at KU Med. These projects demonstrate my proficiency in advance statistics and modeling, and data science.

License

Notifications You must be signed in to change notification settings

BreckEmert/KUMedPortfolio

Repository files navigation

KU Medical Center Portfolio

Hello! I'm Breck Emert, a data science and machine learning student currently completing my Master's in Applied Statistics at KU Med. I'm concurrently researching with the department of biostatistics as the leader of the KU Med AI Research Group, as well as with KU's Brain, Behavior and Quantitative Science program.

This repository showcases my portfolio of advanced projects and analyses completed during my Master's, and demonstrate my proficiency in advanced statistics, modeling, and data science.

Skills Highlighted Here:

  • Data Analysis
  • Machine Learning
  • Programming Languages: Python, R, SAS, JMP

Projects

Below are some of the projects I've completed during my studies.

Categorical Analysis

Logistic Regression Model for Predicting Brain Tumor Malignance

This analysis was completed for my Categorical Analysis course taken in my second semester. The project involved developing a logistic regression model to predict the malignancy of brain tumors (glioblastoma or lower-grade glioma) using data from 857 patients. My key findings include a final model incorporating four significant gene mutations (IDH1, IDH2, TP53, PIK3R1) with a strong predictive power (AUC = 0.90). This study highlights a strong use-case of logistic regression in medical diagnostics and suggests further research into using genetic profiling for early-stage diagnosis.

Business Statistical Analysis

A Statistical Examination of Factors Impacting Market Share in the Packaged Foods Industry

This collaborative project was completed for my Statistical Programming in R course taken in my first semester. Our team aimed to determine the factors that influence market share for an individual project offered by a large packaged goods manufacturer. Our actionable model led to several key recommendations:

  • Leverage demand-invariance during spring: Increase prices when the demand remains steady, maximizing revenue.
  • Optimizing promotional timing: Shift promotions from Spring (where they actually negatively impacted market share) to Winter, where they strongly boost market presence.
  • Adjust Summer pricing: Lower prices in Summer to significantly increase sales volume, outweighing the reduced profit per unit.

Regression Modeling

Housing Affordability in the United States

This project, completed for my Linear Regression course in my first semester, analyzed the impact of population density on rental prices across the United States. The study yielded a clean and interpretable model of rental markets, as a contribution to help back our political conversations with data.

Machine Learning

Brain-Inspired Temporal Attention in Transformers

In-Progress
This paper proposes a novel and interpretable attention mechanism which reduces the computational load of attention. It is inspired by neuroscience and is accompanied by an extensive review of supporting literature. I am currently performing extensive testing of its performance and interpretability using my transformer model as the foundation.

Contact

Please connect with me on LinkedIn or send me an email at breckemert@gmail.com.

About

This repository showcases my portfolio of advanced projects and analyses completed during my Master's in Applied Statistics and Data Science at KU Med. These projects demonstrate my proficiency in advance statistics and modeling, and data science.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published