What is data leakage, and what are real-world examples?
Reviewed by Jason Burns, Editorial Steward · Last updated
Data leakage in machine learning is the accidental inclusion of information in training data that would not be available at prediction time, causing models to look good in evaluation and fail in production.