AUC-ROC Interpretation

Hard

machine-learning Stripe PayPal Capital One

Your binary classifier has an AUC-ROC of 0.92 on a dataset where 95% of samples belong to the positive class. A colleague says the high AUC proves the model is excellent. What is the most valid concern?

Multiple Choice

AUC-ROC is always a reliable metric regardless of class distribution AUC-ROC can be misleadingly optimistic with severe class imbalance; precision-recall AUC is more informative here An AUC of 0.92 means the model correctly classifies 92% of all samples AUC-ROC cannot be computed for imbalanced datasets

Correct!

Incorrect — try again next time!

Upgrade to Pro to unlock solutions and explanations

AUC-ROC Interpretation

Multiple Choice

Related problems

Deep-dive guides