The Case for Evaluating Causal Models Using Interventional Measures and Empirical Data

TL;DR

This paper advocates for evaluating causal models using interventional measures on empirical data, highlighting that current practices rely too much on structural and synthetic data, which limits understanding of model performance.

Contribution

It proposes a shift towards interventional and empirical evaluation methods for causal models, demonstrating their feasibility and importance over traditional approaches.

Findings

Interventional measures differ significantly from structural measures in evaluation results.

Empirical data-based evaluation is feasible with existing datasets.

Current evaluation practices rarely use interventional and empirical methods.

Abstract

Causal inference is central to many areas of artificial intelligence, including complex reasoning, planning, knowledge-base construction, robotics, explanation, and fairness. An active community of researchers develops and enhances algorithms that learn causal models from data, and this work has produced a series of impressive technical advances. However, evaluation techniques for causal modeling algorithms have remained somewhat primitive, limiting what we can learn from experimental studies of algorithm performance, constraining the types of algorithms and model representations that researchers consider, and creating a gap between theory and practice. We argue for more frequent use of evaluation techniques that examine interventional measures rather than structural or observational measures, and that evaluate those measures on empirical data rather than synthetic data. We survey the current practice in evaluation and show that the techniques we recommend are rarely used in practice. We show that such techniques are feasible and that data sets are available to conduct such evaluations. We also show that these techniques produce substantially different results than using structural measures and synthetic data.