A-PETE: Adaptive Prototype Explanations of Tree Ensembles

TL;DR

The paper addresses explainability for tree ensembles by introducing A-PETE, an automatic prototype selection method that constructs a compact set of prototypes using a distance d^{TE} based on leaf co-occurrence across trees. It formulates prototype selection as a medoid-like problem solved via a greedy submodular approach, with a stopping rule that adaptively determines the number of prototypes. Empirical results show A-PETE achieves predictive performance comparable to adaptive Greedy Submodular methods and RF baselines while yielding interpretable prototype sets, supporting both global and local explanations. The work enables automatic, scalable prototype-based interpretation of Random Forests without the need for manual tuning of prototype counts, enhancing practical interpretability of tree ensembles.

Abstract

The need for interpreting machine learning models is addressed through prototype explanations within the context of tree ensembles. An algorithm named Adaptive Prototype Explanations of Tree Ensembles (A-PETE) is proposed to automatise the selection of prototypes for these classifiers. Its unique characteristics is using a specialised distance measure and a modified k-medoid approach. Experiments demonstrated its competitive predictive accuracy with respect to earlier explanation algorithms. It also provides a a sufficient number of prototypes for the purpose of interpreting the random forest classifier.