Problem
Research questions and friction points this paper is trying to address.
Estimates heterogeneous treatment effects interpretably
Forms data-adaptive subgroups using fused lasso
Consistently estimates treatment effects under general conditions
A causal fused lasso for interpretable heterogeneous treatment effects estimation
🤖 AI Summary
This paper addresses the dual challenges of heterogeneous treatment effect (HTE) estimation and interpretable subgroup discovery in causal inference. We propose a data-adaptive method that first orders units by propensity or prognostic scores and matches treated and control units accordingly; then applies fused lasso along this ordered score axis to estimate piecewise-constant conditional average treatment effects (CATEs). To our knowledge, this is the first work to incorporate fused lasso into HTE estimation—eliminating the need for pre-specified subgroup structures and automatically identifying contiguous, semantically meaningful intervals of homogeneous treatment effects. Under general covariate and treatment assignment mechanisms, we establish theoretical consistency guarantees for the CATE estimator. Extensive experiments demonstrate that the method achieves predictive accuracy competitive with state-of-the-art approaches while maintaining strong interpretability through its transparent, interval-based effect representation.
📝 Abstract
We propose a novel method for estimating heterogeneous treatment effects based on the fused lasso. By first ordering samples based on the propensity or prognostic score, we match units from the treatment and control groups. We then run the fused lasso to obtain piecewise constant treatment effects with respect to the ordering defined by the score. Similar to the existing methods based on discretizing the score, our methods yield interpretable subgroup effects. However, existing methods fixed the subgroup a priori, but our causal fused lasso forms data-adaptive subgroups. We show that the estimator consistently estimates the treatment effects conditional on the score under very general conditions on the covariates and treatment. We demonstrate the performance of our procedure using extensive experiments that show that it can be interpretable and competitive with state-of-the-art methods.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Fused lasso for heterogeneous treatment effects
Data-adaptive subgroups via propensity scores
Piecewise constant interpretable subgroup effects
🔎 Similar Papers
No similar papers found.
