ROC vs. AUC

[SaVoAMP]

Hey,

I was trying to find anomalies using IForest. Is there a particular reason why the evaluate_print function shows a ROC score? It's a bit confusing to me since the receiver operating characteristic (ROC) refers to a curve that plots the recall over the false positive rate and therefore cannot be described by a scalar value. I guess you are referring to the area under curve (AUC)? If that is the case, I would suggest calling it AUC instead in order to avoid confusion. What are your thoughts on that?

[yzhao062]

Thanks for checking. it is a common practice to call ROC-AUC as ROC in ML research :)