I have two classes and, in total, 40 subjects
(20,20). I used an SVM to implement classification and did a leave-one-out cross validation to get an overall performance of the classifier.
How could I draw an ROC curve for this situation?
Is it even necessary to draw a ROC curve in this situation?