Reproduce “SelectArena: A Modern Benchmark for Feature Selection on Real-world Tabular Data” 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
Bastian Schäfer, Dominika Matus, Lennart Purucker, Ricardo Knauer, Erik Rodner, Frank Hutter, Anton Sokolov, Arnaud Liefooghe
SelectArena compares 15 feature-selection approaches, including random selection, on 71 curated real-world tabular datasets using five downstream predictors. It measures predictive performance, runtime, stability, and validity under fixed-cardinality selections. RFImportance averages 26.23% improvement over random at 45.19 seconds per 1,000 features; F-Test averages 22.76% at 0.79 seconds and leads stability and validity under the benchmark’s one-hour limit and noise test.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.