Counterfactual Metarules for Local and Global Recourse

TL;DR

T-CREx is a new model-agnostic method that generates human-readable counterfactual explanations for individuals and groups, using tree-based surrogate models to provide diverse recourse options and analyze model behavior globally.

Contribution

It introduces T-CREx, a novel approach combining metarules and tree-based surrogates for efficient, interpretable local and global counterfactual explanations.

Findings

T-CREx outperforms existing rule-based methods in aggregate performance.

It is significantly faster than previous approaches.

Provides both local recourse options and global model insights.

Abstract

We introduce T-CREx, a novel model-agnostic method for local and global counterfactual explanation (CE), which summarises recourse options for both individuals and groups in the form of human-readable rules. It leverages tree-based surrogate models to learn the counterfactual rules, alongside 'metarules' denoting their regions of optimality, providing both a global analysis of model behaviour and diverse recourse options for users. Experiments indicate that T-CREx achieves superior aggregate performance over existing rule-based baselines on a range of CE desiderata, while being orders of magnitude faster to run.