TraCE: Trajectory Counterfactual Explanation Scores

TL;DR

TraCE introduces a model-agnostic framework to evaluate progress in sequential decision making using counterfactual explanations, demonstrated across healthcare and climate change domains.

Contribution

The paper presents TraCE scores, a novel modular method to quantify progress in complex sequential decision tasks through counterfactuals.

Findings

TraCE effectively summarizes progress in healthcare decision scenarios.

TraCE scores are applicable to climate change policy evaluation.

The framework is model-agnostic and adaptable across domains.

Abstract

Counterfactual explanations, and their associated algorithmic recourse, are typically leveraged to understand, explain, and potentially alter a prediction coming from a black-box classifier. In this paper, we propose to extend the use of counterfactuals to evaluate progress in sequential decision making tasks. To this end, we introduce a model-agnostic modular framework, TraCE (Trajectory Counterfactual Explanation) scores, which is able to distill and condense progress in highly complex scenarios into a single value. We demonstrate TraCE's utility across domains by showcasing its main properties in two case studies spanning healthcare and climate change.