AI Glossary

Data leakage

What does “Data leakage” mean?

Data leakage: Allowing information that would be unavailable in deployment to influence training or evaluation.

Data leakage visual guide: Allowing information that would be unavailable in deployment to influence training or evaluation.
Visual guide Open full size (new tab)

What is an example of “Data leakage”?

A model uses a field recorded only after the outcome it is supposed to predict.

Keep learning

See also

EvalsTermBenchmarkTermValidation and test setsTermCross-validationTermStratified samplingTerm

Same topic

Covariate shiftTermDistribution shiftTermData poisoningTermData exfiltrationTermData privacyTerm