On the Trustworthiness of Tree Ensemble Explainability Methods

TL;DR

This paper evaluates the accuracy and stability of global feature importance methods for tree ensemble models, revealing their limitations in noisy and high-dimensional settings through comprehensive experiments.

Contribution

It provides the first systematic comparison of global feature importance methods' accuracy and stability, highlighting their limitations in real-world scenarios.

Findings

Global feature importance methods lack accuracy with noisy inputs.

These methods are unstable under increased input dimension and data noise.

Model perturbations due to different initializations affect explanation stability.

Abstract

The recent increase in the deployment of machine learning models in critical domains such as healthcare, criminal justice, and finance has highlighted the need for trustworthy methods that can explain these models to stakeholders. Feature importance methods (e.g. gain and SHAP) are among the most popular explainability methods used to address this need. For any explainability technique to be trustworthy and meaningful, it has to provide an explanation that is accurate and stable. Although the stability of local feature importance methods (explaining individual predictions) has been studied before, there is yet a knowledge gap about the stability of global features importance methods (explanations for the whole model). Additionally, there is no study that evaluates and compares the accuracy of global feature importance methods with respect to feature ordering. In this paper, we evaluate the accuracy and stability of global feature importance methods through comprehensive experiments done on simulations as well as four real-world datasets. We focus on tree-based ensemble methods as they are used widely in industry and measure the accuracy and stability of explanations under two scenarios: 1) when inputs are perturbed 2) when models are perturbed. Our findings provide a comparison of these methods under a variety of settings and shed light on the limitations of global feature importance methods by indicating their lack of accuracy with and without noisy inputs, as well as their lack of stability with respect to: 1) increase in input dimension or noise in the data; 2) perturbations in models initialized by different random seeds or hyperparameter settings.