sepplotlib

Separation plots for binary classification problems.

Credits

The one-dimensional separation plot is adapted from code originally produced by Brian Greenhill, Michael D. Ward, and Audrey Sacks. The bi-separation plot and model criticism plot are adapted from code originally produced by Michael Colaresi and Zuhaib Mahmood.

Installation

pip install sepplotlib to install.

Example usage

Please see the accompanied notebook for an example using mock data.

The included figures are objects that expect a pandas DataFrame and strings for the relevant columns. To generate a one-dimensional separation plot for instance, simply run:

import sepplotlib as spl
spl.SeparationPlot(
    df=df,
    y_true="y_true",
    y_pred="y_pred",
    title="Example"
)

Similarly to generate a model criticism plot:

import sepplotlib as spl
spl.ModelCriticismPlot(
    df=df,
    y_true="y_true",
    y_pred="y_pred",
    lab="lab",
    title="Example"
)

And finally, to generate a two-dimensional, bi-separation plot:

import sepplotlib as spl
spl.BiseparationPlot(
    df=df,
    x="y_pred_a",
    y="y_pred_b",
    obs="y_true",
    lab="lab",
    title="Example",
)

Please run help on any of these classes to learn what can be customized (e.g. help(spl.SeparationPlot)).