TabPrep: Closing the Feature Engineering Gap in Tabular Benchmarks
Researchers have introduced TabPrep, a new preprocessing pipeline designed to address the gap in feature engineering for tabular machine learning benchmarks. This pipeline incorporates feature generators targeting specific data patterns, revealing limitations in current model architectures. Integrating TabPrep has demonstrated consistent performance improvements across various model types, often exceeding gains from model-centric advancements alone. AI
IMPACT Improves evaluation of tabular models by integrating feature engineering into benchmarks.