mixmo.core.metrics_ensemble

Ensembling metrics Inspired from * https://github.com/kbogas/EnsembleDiversityTests/blob/master/EnsembleDiversityTests.py * https://github.com/scikit-learn-contrib/DESlib

Functions

compute_mean_without_diagonal(matrix)
compute_pairwise_diversity(targets, …)	Computes the pairwise diversity matrix.
get_accuracy_multi(predictions, y_true)
ratio_errors(y, y_pred1, y_pred2)	Calculates Ratio of errors diversity measure between a pair of classifiers.

Classes

MetricsEnsemble(predictions, names, true)

Class Wrapper to get Diversity Measures over collection of predictions. :param @predictions: list of lists. Each sublist contains the predictions of a classifier :param @names: list of strings. Each string is the name of the classifier. :param @true: list of labels. Each label is the truth label.