Hands-On Artificial Intelligence for Cybersecurity

ROC curve and AUC measure

Among the techniques most commonly used to compare the performance of different classifiers, we have the receiving operating characteristic (ROC) curve, which describes the relationship between the TPR, or sensitivity, and the FPR, or 1 – Specificity, associated with each classifier:

(Image Credits: http://scikit-learn.org/)

How can the performances of the different classifiers be compared?

Let's start by taking into consideration the characteristics that the best possible classifier should have. Its curve would correspond to the pair of values, ​​x = 0 and y = 1, in the ROC space. In other words, the best possible ...