How to assess quality and correctness of classification models? Part 4 - ROC Curve

We test the classifier for different alpha thresholds. Recall that alpha is the threshold of the estimated probability, above which an observation is assigned to one category (positive class) and below to the other category (negative class).