medium10 pts · first answer countsmodel-evaluationclassification-metricsimbalanced-data
ROC-AUC looks great, alerts are useless
A fraud model has ROC-AUC of 0.97, but fraud is 0.1% of transactions and analysts say most alerts are false. Which metric better shows this?
Discussion 0
Join free to post in the discussion.
Related questions
- Big gap between train and validation accuracyMachine Learning Engineer
- Why keep a separate test setMachine Learning Engineer
- Definition of precisionMachine Learning Engineer
- Accuracy on imbalanced dataMachine Learning Engineer
- Recall from confusion countsMachine Learning Engineer
- How often a row is validated in k-foldMachine Learning Engineer
- Why stratify the foldsMachine Learning Engineer
- Specificity by handMachine Learning Engineer