Tree Space Prototypes: Another Look at Making Tree Ensembles Interpretable

TL;DR

This paper introduces a prototype-based interpretability method for tree ensembles, surfacing representative points per class, which can match or outperform the original models and improve human understanding over feature attribution methods.

Contribution

It proposes a novel prototype selection approach with theoretical guarantees for tree ensemble interpretability, including a new distance for Gradient Boosted Trees and adaptive prototype selection.

Findings

Prototypes can match or outperform original tree ensembles as classifiers.

Humans predict ensemble outputs better with prototypes than with Shapley values.

The method provides flexible, class-specific prototype counts with theoretical backing.

Abstract

Ensembles of decision trees perform well on many problems, but are not interpretable. In contrast to existing approaches in interpretability that focus on explaining relationships between features and predictions, we propose an alternative approach to interpret tree ensemble classifiers by surfacing representative points for each class -- prototypes. We introduce a new distance for Gradient Boosted Tree models, and propose new, adaptive prototype selection methods with theoretical guarantees, with the flexibility to choose a different number of prototypes in each class. We demonstrate our methods on random forests and gradient boosted trees, showing that the prototypes can perform as well as or even better than the original tree ensemble when used as a nearest-prototype classifier. In a user study, humans were better at predicting the output of a tree ensemble classifier when using prototypes than when using Shapley values, a popular feature attribution method. Hence, prototypes present a viable alternative to feature-based explanations for tree ensembles.