lemma-leakage

name	lemma-leakage
description	Audit suspicious model performance or a pipeline for target, preprocessing, temporal, group, sampling, or duplicate contamination.

Leakage audit

Deliver

Rank leakage findings by severity and show reproducible before/after scores, the contaminated path, the corrected validation design, and remaining risk.

Check

• Reconstruct information available at the real prediction moment.
• Audit features and missingness for target proxies and post-outcome facts.
• Verify that imputers, encoders, selection, and scaling are fit within each training fold.
• Check time travel, repeated entities, near duplicates, source overlap, and outcome-dependent sampling.
• Remove suspicious features or strengthen the split and compare with a simple baseline.

Do not rationalize an implausible score, rely only on feature importance, randomly split temporal or repeated-entity data, or continue tuning before material leakage is resolved.

For the full leakage taxonomy, read references/deep-guide.md.