Reproduce “Optimal Subgroup Discovery at Every Support Threshold” 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

Lincen Yang, Qi Huang, Niki van Stein, Thomas Bäck, Matthijs van Leeuwen, Zhong Li

AtomSD finds subgroups whose target distributions differ from the population while meeting a user-selected minimum support. It learns regions with similar conditional target distributions and searches combinations of whole regions plus at most one partial region, under a partition-based model. Across 19 benchmarks, it averaged rank 1.25 and won 78 of 95 comparisons; this does not guarantee global optimality.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.