How does ROC-AUC work on data with a strong class imbalance?

ROC-AUC (Area Under the ROC Curve) measures a model's ability to distinguish between classes by evaluating the ratio of true positive to false positive results at different thresholds. In cases of strong class imbalance, ROC-AUC can be less informative because it does not account for the absolute number of errors, only the relative metrics. In such cases, a model may show a high ROC-AUC even if it poorly recognizes the minority class. Therefore, for imbalanced data, additional metrics sensitive to class imbalance, such as Precision-Recall AUC or F1 score, are often used.