Daze¶
The sklearn.metrics
module allows for the plotting of a confusion matrix from a
classifier (with sklearn.metrics.plot_confusion_matrix()
, or directly from a pre-computed confusion matrix (with the internal sklearn.metrics.ConfusionMatrixDisplay
class).
A confusion matrix shows the discrepancy between the true labels of a dataset and the labels predicted by a classifier.
While the confusion matrix plots generated by Scikit-Learn are very informative, they omit important evaluation measures that can summarize classification performance. True positives, precision, F1 score and accuracy are example of such measures – all of which can be derived from the confusion matrix. The sklearn.metrics.classification_report()
function in the same module provides these measures.
Daze adjusts plot_confusion_matrix
to incorporate these evaluation measures directly in the confusion matrix plot, while still maintaining a very similar API
to the original Scikit-Learn function.