Consistent Individualized Feature Attribution for Tree Ensembles

TL;DR

This paper introduces a fast, exact method for computing consistent, individualized feature attributions in tree ensemble models using SHAP values, improving interpretability and clustering of features.

Contribution

The paper develops a novel, efficient algorithm for exact SHAP value computation in tree ensembles, ensuring consistency and extending to interaction effects.

Findings

Exact SHAP values improve interpretability.

Enhanced clustering based on feature attributions.

Better alignment with human intuition.

Abstract

Interpreting predictions from tree ensemble methods such as gradient boosting machines and random forests is important, yet feature attribution for trees is often heuristic and not individualized for each prediction. Here we show that popular feature attribution methods are inconsistent, meaning they can lower a feature's assigned importance when the true impact of that feature actually increases. This is a fundamental problem that casts doubt on any comparison between features. To address it we turn to recent applications of game theory and develop fast exact tree solutions for SHAP (SHapley Additive exPlanation) values, which are the unique consistent and locally accurate attribution values. We then extend SHAP values to interaction effects and define SHAP interaction values. We propose a rich visualization of individualized feature attributions that improves over classic attribution summaries and partial dependence plots, and a unique "supervised" clustering (clustering based on feature attributions). We demonstrate better agreement with human intuition through a user study, exponential improvements in run time, improved clustering performance, and better identification of influential features. An implementation of our algorithm has also been merged into XGBoost and LightGBM, see http://github.com/slundberg/shap for details.