Statistical Significance of Feature Importance Rankings

TL;DR

This paper introduces statistically rigorous methods to verify and identify the most important features in machine learning models, ensuring high-probability correctness despite sampling variability.

Contribution

It proposes hypothesis testing-based techniques and sampling algorithms to reliably determine feature importance rankings with theoretical guarantees.

Findings

Algorithms accurately identify top features with high probability

Methods validated empirically on SHAP and LIME importance scores

Provides stability assessment for feature importance rankings

Abstract

Feature importance scores are ubiquitous tools for understanding the predictions of machine learning models. However, many popular attribution methods suffer from high instability due to random sampling. Leveraging novel ideas from hypothesis testing, we devise techniques that ensure the most important features are correct with high-probability guarantees. These assess the set of $K$ top-ranked features, as well as the order of its elements. Given a set of local or global importance scores, we demonstrate how to retrospectively verify the stability of the highest ranks. We then introduce two efficient sampling algorithms that identify the $K$ most important features, perhaps in order, with probability exceeding $1-\alpha$. The theoretical justification for these procedures is validated empirically on SHAP and LIME.