easy5 pts · first answer countsclassification-metricsimbalanced-data
Accuracy on imbalanced data
Only 1% of transactions are fraud. A model that always predicts "not fraud" gets 99% accuracy. Which metric would better show that this model is useless?
Discussion 0
Join free to post in the discussion.
Related questions
- Definition of precisionMachine Learning Engineer
- Recall from confusion countsMachine Learning Engineer
- Handling class imbalanceMachine Learning Engineer
- F1 score from a confusion matrixMachine Learning Engineer
- Log loss of a poor predictionMachine Learning Engineer
- AUC from four scoresMachine Learning Engineer
- ROC-AUC after calibrationMachine Learning Engineer
- Precision from a confusion matrixData Scientist