Regional Explanations: Bridging Local and Global Variable Importance

TL;DR

This paper critically examines local attribution methods like LIME and Local Shapley Values, revealing their limitations and proposing R-LOCO, a regional approach that improves local explanation fidelity by integrating global methods within input space segments.

Contribution

It identifies fundamental flaws in existing local attribution methods and introduces R-LOCO, a novel regional explanation technique that enhances local attribution accuracy and stability.

Findings

LIME and Local Shapley Values can incorrectly assign importance to irrelevant features.

R-LOCO segments input space to improve local attribution fidelity.

Regional explanations provide more stable and accurate feature importance.

Abstract

We analyze two widely used local attribution methods, Local Shapley Values and LIME, which aim to quantify the contribution of a feature value $x_i$ to a specific prediction $f(x_1, \dots, x_p)$. Despite their widespread use, we identify fundamental limitations in their ability to reliably detect locally important features, even under ideal conditions with exact computations and independent features. We argue that a sound local attribution method should not assign importance to features that neither influence the model output (e.g., features with zero coefficients in a linear model) nor exhibit statistical dependence with functionality-relevant features. We demonstrate that both Local SV and LIME violate this fundamental principle. To address this, we propose R-LOCO (Regional Leave Out COvariates), which bridges the gap between local and global explanations and provides more accurate attributions. R-LOCO segments the input space into regions with similar feature importance characteristics. It then applies global attribution methods within these regions, deriving an instance's feature contributions from its regional membership. This approach delivers more faithful local attributions while avoiding local explanation instability and preserving instance-specific detail often lost in global methods.