Skip to content

Consistent Individualized Feature Attribution for Tree Ensembles

Authors: Scott M. Lundberg, Gabriel G. Erion, Su-In Lee

Published: 2018 (Preprint)

Source: arXiv

Algorithm: TreeSHAP

arXiv: 1802.03888

Summary

Derives a polynomial-time exact algorithm for computing Shapley values on tree ensembles (TreeSHAP), exploiting the tree structure to avoid exponential enumeration of coalition orderings. Demonstrates that common tree feature importance heuristics are inconsistent, while TreeSHAP is the unique consistent and locally accurate attributor for models like XGBoost and Random Forests.

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.

Primary

Paper arxiv.org

Standard

arXiv Abstract 1802.03888 arXiv PDF 1802.03888 arXiv HTML 1802.03888

Alternate

scottlundberg.com scottlundberg.com GitHub: shap/shap github.com

Tags

  • Explainability

  • Feature attribution

  • SHAP

  • Tree ensembles

  • TreeSHAP

  • Gradient boosting

  • Random forests

  • Shapley values