Feature Synergy, Redundancy, and Independence in Global Model Explanations using SHAP Vector Decomposition
This provides a more detailed global explanation method for feature interactions in machine learning models, though it appears incremental as it builds upon existing SHAP frameworks.
The paper tackles the problem of explaining pairwise feature dependencies in supervised models by introducing a new formalism that decomposes SHAP values into synergistic, redundant, and independent components, and demonstrates its utility on a constructed dataset.
We offer a new formalism for global explanations of pairwise feature dependencies and interactions in supervised models. Building upon SHAP values and SHAP interaction values, our approach decomposes feature contributions into synergistic, redundant and independent components (S-R-I decomposition of SHAP vectors). We propose a geometric interpretation of the components and formally prove its basic properties. Finally, we demonstrate the utility of synergy, redundancy and independence by applying them to a constructed data set and model.