balnet: Pathwise Estimation of Covariate Balancing Propensity Scores

2026-02-20

AI Summary

This study addresses the scalability challenges of covariate balancing and propensity score estimation in high-dimensional or complex data, such as satellite imagery. The authors propose a pathwise estimation method based on a logistic covariate balancing loss, which for the first time integrates Yang and Hastie’s (2024) general elastic net solver into covariate-balancing propensity score computation. This approach accommodates nonsmooth, grouped, and feature-specific penalties, efficiently generating a continuous solution path from maximal imbalance to a target balance level. The method is applied to a pixel-level spatial synthetic control task using wildfire satellite images, where it successfully constructs treatment effect weights, demonstrating its effectiveness and flexibility in real-world, high-dimensional settings.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Search and Optimization: Mixed Discrete/Continuous Search
• Reasoning under Uncertainty: Probabilistic Programming

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

We present balnet, an R package for scalable pathwise estimation of covariate balancing propensity scores via logistic covariate balancing loss functions. Regularization paths are computed with Yang and Hastie (2024)'s generic elastic net solver, supporting convex losses with non-smooth penalties, as well as group penalties and feature-specific penalty factors. For lasso penalization, balnet computes a regularized balance path from the largest observed covariate imbalance to a user-specified fraction of this maximum. We illustrate the method with an application to spatial pixel-level balancing for constructing synthetic control weights for the average treatment effect on the treated, using satellite data on wildfires.

Problem

• covariate balancing
• propensity scores
• causal inference
• regularization path
• synthetic control

Innovation

• covariate balancing
• propensity scores
• elastic net
• regularization path
• synthetic control