Glossary term
Glossary term
Evaluation and Benchmarks
A curve of precision versus recall at different classification thresholds.
Created for this library
A fraud team uses the precision-recall curve to choose an operating threshold that balances customer friction against fraud losses.
A medical screening team uses the precision-recall curve when evaluating rare-disease detectors because it is more informative than ROC for class imbalance.
A retention team uses the precision-recall curve to pick the threshold that fills the call center's daily capacity.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License