ML//Evaluation//PR-AUC

Area under the precision-recall curve. The evaluation metric that actually respects rare positives.

Area under the precision-recall curve. The evaluation metric that actually respects rare positives.

Unlike ROC-AUC, it is sensitive to class imbalance: on a rare-event problem a high ROC-AUC can hide a useless precision.

The baseline equals the positive rate, not 0.5, so "PR-AUC 0.1" can still be real signal when positives are 1% of the data.

Use it whenever the positive class is scarce and false alarms are costly.