AUC-ROC
AUC-ROC measures how well a classifier separates two classes across every possible threshold, summarized as one number from 0.5 (random) to 1.0 (perfect). Unlike accuracy, it doesn't depend on where you set the decision cutoff.
Also known as: AUC, ROC AUC, area under the curve
The ROC curve plots a classifier’s true-positive rate against its false-positive rate as you sweep the decision threshold from strict to lenient. AUC-ROC is the area under that curve: it captures, in a single number, how well the model ranks a random positive above a random negative. 0.5 is coin-flip; 1.0 is perfect separation.
Its strength is that it’s threshold-independent — it judges the model’s underlying ability to discriminate, not the cutoff you happened to pick. The caveat: on heavily imbalanced data, AUC-ROC can look flattering because false positives are diluted by a large negative class. When positives are rare and costly to miss, a precision-recall curve often tells the more honest story.