Shapley variable importance cloud for machine learning models

TL;DR

This paper extends the Shapley variable importance cloud (ShapleyVIC) method to machine learning models, providing a robust, model-agnostic approach for variable importance assessment with uncertainty quantification, complementing existing SHAP analyses.

Contribution

The work broadens ShapleyVIC's application from regression to general machine learning models, enhancing interpretability and trustworthiness of black-box models.

Findings

ShapleyVIC offers comprehensive variable importance insights.

It provides uncertainty intervals for variable contributions.

The method is applicable across various machine learning models.

Abstract

Current practice in interpretable machine learning often focuses on explaining the final model trained from data, e.g., by using the Shapley additive explanations (SHAP) method. The recently developed Shapley variable importance cloud (ShapleyVIC) extends the current practice to a group of "nearly optimal models" to provide comprehensive and robust variable importance assessments, with estimated uncertainty intervals for a more complete understanding of variable contributions to predictions. ShapleyVIC was initially developed for applications with traditional regression models, and the benefits of ShapleyVIC inference have been demonstrated in real-life prediction tasks using the logistic regression model. However, as a model-agnostic approach, ShapleyVIC application is not limited to such scenarios. In this work, we extend ShapleyVIC implementation for machine learning models to enable wider applications, and propose it as a useful complement to the current SHAP analysis to enable more trustworthy applications of these black-box models.