Reproduce “Out-of-sample scoring and automatic selection of causal estimators” 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
Egor Kraev, Timo Flesch, Hudson Taylor Lekunze, Mark Harley, Pere Planell Morell
The authors introduce scores for selecting and tuning causal estimators when individual treatment effects are unobservable. They evaluate models using held-out patterns implied by the experiment or causal setup, rather than comparing predictions with unknown individual effects. In synthetic randomized-treatment CATE experiments, all three scores guide selection toward lower true effect-estimation error; performance is weaker with observed confounders.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.