AI engineering glossary · ML fundamentals
What is target leakage?
Target leakage (also called Label leakage): A form of data leakage in which a feature is directly or indirectly produced by the outcome being predicted, or recorded after it. The model then learns a shortcut that is impossible to use when making genuine predictions.
Target leakage: an example
Predicting hospital readmission with a column 'discharge medication changed after readmission' gives the answer away.
Related terms
More ML fundamentals terms
- Temporal leakage
- Test set
- Time-series split
- Train-test contamination
- Training
- Training error
- Training example
- Training set
Where you learn it
Target leakage is part of the AI Engineering Program, which teaches it with notes, a hands-on lab and an exam. The first lesson is free.