ROC AUC
Also: area under the ROC curve
The probability that a randomly chosen positive example is scored above a randomly chosen negative one. It measures ranking quality across all thresholds, independent of where you eventually set one.
A definition is not an understanding. This is taught properly in Choosing the metric that matches the cost of being wrong, part of Model evaluation & metrics.
Learn this properlyTerms from the same topic