Version 0.6#

This release includes an enhancement mainly in the MultiViewDecisionTreeClassifier and HonestForestClassifier, and a new generative model for the make_trunk_classification.

Changelog#

Enhancement sktree.tree.MultiViewDecisionTreeClassifier now rounds up the number of features to split on to the nearest integer when applying max_features to each feature view, by Adam Li (##183).
Feature sktree.tree.MultiViewDecisionTreeClassifier now supports an array passed in for max_features, which applies a different max_features argument per view, by Adam Li (##183).
Fix sktree.tree.MultiViewDecisionTreeClassifier now correctly handles the case where there is one feature view that is exhausted, and another that is not for apply_max_features_per_feature_set = False, by Adam Li (##183).
Fix sktree.stats.FeatureImportanceForestClassifier now correctly passes metric kwargs to the null distribution function, by Adam Li (##183).
Enhancement sktree.datasets.make_trunk_classification() now has a generative model based on Trunk and banded covariance, sktree.datasets.approximate_clf_mutual_information() and sktree.datasets.approximate_clf_mutual_information_with_monte_carlo() to approximate mutual information either numerically or via Monte-Carlo, by Adam Li and Haoyin Xu (##199).
Enhancement sktree.HonestForestClassifier now has a fitted property oob_samples_, which reproduces the sample indices per tree that is out of bag, by Adam Li (##200).
Enhancement sktree.HonestForestClassifier will allow one to bootstrap sample higher than the number of samples, controlled by the max_samples keyword argument by Adam Li (##206).
Feature sktree.HonestForestClassifier now allows one to specify the number of sub-samples to use for the honest trees without having to bootstrap sample. This is specified by the max_samples parameter. By Adam Li (##210)

Code and Documentation Contributors#

Thanks to everyone who has contributed to the maintenance and improvement of the project since version inception, including: