By Reiichiro Nakano.
There are a number of visualizations that frequently pop up in machine learning. Scikit-plot is a humble attempt to provide aesthetically-challenged programmers (such as myself) the opportunity to generate quick and beautiful graphs and plots with as little boilerplate as possible.
Here's a quick example to generate the precision-recall curves of a Keras classifier on a sample dataset:
# Import what's needed for the Functions API
import matplotlib.pyplot as plt
import scikitplot.plotters as skplt
# This is a Keras classifier. We'll generate probabilities on the test set.
keras_clf.fit(X_train, y_train, batch_size=64, nb_epoch=10, verbose=2)
probas = keras_clf.predict_proba(X_test, batch_size=64)
# Now plot.
Then just run:
pip install scikit-plot
Or if you want, clone this repo and run
python setup.py install
at the root folder.
To access the full resource on Github, with numerous examples and pieces of code, click here.
Top DSC Resources