X-SHAP: towards multiplicative explainability of Machine Learning

TL;DR

X-SHAP is a novel, model-agnostic method that extends SHAP to assess multiplicative contributions of variables, enhancing interpretability for models with multiplicative interactions in fields like insurance and biology.

Contribution

The paper introduces X-SHAP, a new approach that extends additive SHAP to evaluate multiplicative effects, providing a more accurate interpretability tool for certain models.

Findings

X-SHAP effectively captures multiplicative feature interactions.

The method improves interpretability in sectors using Generalized Linear Models.

Comparison shows X-SHAP outperforms traditional techniques in identifying feature importance.

Abstract

This paper introduces X-SHAP, a model-agnostic method that assesses multiplicative contributions of variables for both local and global predictions. This method theoretically and operationally extends the so-called additive SHAP approach. It proves useful underlying multiplicative interactions of factors, typically arising in sectors where Generalized Linear Models are traditionally used, such as in insurance or biology. We test the method on various datasets and propose a set of techniques based on individual X-SHAP contributions to build aggregated multiplicative contributions and to capture multiplicative feature importance, that we compare to traditional techniques.