ROC curves for the automatically generated classifier codes

Hi everyone,

I've generated some code for several classiffiers using the Classiffication Learner app. These codes only give the accuracy to validate the classifiers, but they don't give any ROC curve values. I want to add some code to compute the AUC of the ROC curves, but I'm a bit confused.

I'm using the perfcurve function, so I have to give it the actual values of the labels and the classification scores. I do have the values of the labels, but for the scores, I have a reduced set of values as the classifier is being "partitioned" in K folds.

Can someone guide me on how should I compute the ROC values in this case?

Thank you!!