ROC-AUC on võrdne 0,8-ga. Kuidas see muutub, kui korrutame skoori 1,25-ga?

ROC-AUC (Area Under the Receiver Operating Characteristic Curve) measures the quality of the model's ranking, that is, how well the model distinguishes between positive and negative classes. If all predicted probabilities (scores) are multiplied by 1.25, the relative order of predictions does not change, as multiplication by a positive number is a monotonic operation.

Since ROC-AUC depends only on the order of predictions and not on their absolute values, the ROC-AUC value will remain the same — 0.8.