Dear experts,
I’m using TMVA for training a BDTs classifier model. I noticed that after the model is ready, I can request a ROC curve of the BDTs trained and TMVA will tell me its AUC. But now I would like to know the discriminative power of the variables entered for training. Is there a method that I can use via TMVA to request the ROC curve of a variable as a TGraph object?
I setup via adding a signal tree and a background tree.