Reproduce “Conformal Counterfactual Inference under Hidden Confounding” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Zonghao Chen, Ruocheng Guo, Jean-François Ton, Yang Liu
The authors present conformal intervals for counterfactual outcomes with marginal coverage despite unobserved confounding. The method estimates observational-to-interventional joint density ratios and weights calibration errors. On Yahoo!R3, exact wSCP-DR achieves coverage 0.952 ± 0.001 and mean width 5.140 ± 0.001, versus Naive at 0.899 ± 0.001 coverage and 6.047 ± 0.001 width. The guarantee is marginal; Inexact lacks it.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.