Analyze the Performance of an Image Classifier

ClassifierMeasurements allows the measurement of a variety of performance metrics for a classifier. This example demonstrates some of its capabilities on an image classifier.

Train a classifier to identify trees from images.

Obtain a performance report for the classifier using a test set.

Find the two best classified test examples and check their classifications.

Extract the two worst classified test examples and check their classifications.

Obtain all test examples of eucalyptuses misclassified as oaks.

Compute the micro-averaged F1 score and Matthews correlation coefficient, along with their uncertainties.