A binary classifier's ROC curve has an Area Under the Curve (AU

AUC (Area Under the ROC Curve) measures a classifier's ability to rank a randomly chosen positive instance higher than a randomly chosen negative instance, across all possible classification thresholds. An AUC of 0.5 corresponds to the ROC curve lying exactly on the diagonal line from (0,0) to (1,1), which is precisely what a classifier making random predictions (no real discriminative power between classes) would produce — regardless of the specific accuracy value or class balance, which are separate, threshold-dependent metrics.