Per Category Modeling Backfire

Per-category/per-family models can produce WORSE leaderboard scores than a single global model, even when per-category CV is better. Use when: (1) Considering training separate models per product category/family/store in tabular competitions, (2) Per-category CV improves but LB degrades, (3) Each category has <100K rows from a larger dataset. Covers data volume thresholds, hybrid fallback strategies, and when per-category modeling is appropriate vs counterproductive.

topprismdata Updated 5 repo stars
TopPrism dual-prism visual

File contents

topprismdata/cultivating-ml-agent commit b0ca55cd25

Frequently asked questions

npx get-skill add topprismdata/per-category-modeling-backfire