Downsampled plotting in Matplotlib

An axes class enabling downsampled plotting in Matplotlib

Import the module by

import dsaxes

Line objects in matplotlib.pyplot.plots or its derivatives are downsampled for better responsivity with xy-plots of many data points.

dsaxes dynamically overloads the plot and legend commands of matplotlib

With dsaxes enabled, the toolbar of figures generated has a button which toggles downsampling on and off. Furthermore, legends are interactive, i.e.

• hovering the mouse cursor over a line indicator in the legend brings the associated line to front (sometimes it is a bit hard to hit the line with the cursor, just try around a bit).
• clicking on a line indicator in the legend toggles the visibility of the associated line

The dsaxes behaviour can be monitored and switched between normal axes Axes.plot behaviour (non-downsampling plots, non-interactive legends) and dsaxes behaviour (downsampling plots, interactive legends) by using the switch class, see this sample code:

    import matplotlib.pyplot as plt
    import numpy as np
    import pandas as pd
    import dsaxes # import the dsaxes module

    ds = dsaxes.switch() # instantiate an instance of the switch class
    ds_state = ds.get_state() # save initial state for reconstruction later

    ds.set_state(True) # switch on dsaxes behaviour
    print(ds.get_state()) # prints "True", meaning dsaxes behaviour is enabled
    ds.set_state() # toggle state , here changes from dsaxes to axes.Axes behaviour
    print(ds.get_state()) # prints "False", meaning axes.Axes behaviour is enabled
    ds.set_state() # toggle state back to dsaxes behaviour
    print(ds.get_state()) # prints "True", meaning dsaxes behaviour is enabled

    # generate some test data:
    N=1e6
    t= np.arange(N)/N
    y= np.random.randn(int(N))
    y[3]=np.nan
    y1=pd.DataFrame(index= pd.date_range('2019',periods=N, freq='S'), data=y)

    # plot pandas dataframe converted to numpy arrays:
    plt.close(fig=1)
    plt.figure(1)
    ax = plt.axes()
    ax.plot(y1.index, y1.values,'o',label='1') 
    ax.plot(y1.index, y1.values, linestyle= '-',label='2') 
    ax.legend(interactive=False) # switch off interacitity for this legend (default = True)

    # plot a pandas dataframe (auto-generates a legend)
    plt.close(fig=2)
    plt.figure(2)
    ax = plt.axes()
    y1.plot(ax=ax, marker='o', linestyle= 'None') 
    y1.plot(ax=ax, linestyle= '-') 

    ds.set_state() # switch off dsaxes behaviour, enable normal axes.Axes behaviour

    # plot numpy array
    plt.close(fig=3)
    plt.figure(3)
    ax = plt.axes()
    plt.plot(t,y,'o')
    plt.plot(t,y,'-')
    ax.legend(['1','2']) # if dsaxes behaviour is switched off, legend must not have the "interactive" argument, as that would result in an error.

    ds.set_state(ds_state) # reset state to state saved at beginning of script