ROC AUC plots for multi-labels

1

Hello, I want to do roc auc plot for every class in a problem where the labels are: 0, 1, 2, 3, 4 ,5 ,6.
I found this code:

                    label_binarize(YTest, classes=[0, 1, 2, 3, 4, 5, 6])
                    n_classes = YTest.shape[1]
                    classifier = OneVsRestClassifier(
                        svm.SVC(kernel="linear", probability=True, random_state=random_state)
                    )
                    y_score = classifier.fit(XTest, YTest).decision_function(XTest)
                    fpr = dict()
                    tpr = dict()
                    roc_auc = dict()
                    for i in range(n_classes):
                        fpr[i], tpr[i], _ = roc_curve(YTest[:, i], y_score[:, i])
                        roc_auc[i] = auc(fpr[i], tpr[i])
                    fpr["micro"], tpr["micro"], _ = roc_curve(YTest.ravel(), y_score.ravel())
                    roc_auc["micro"] = auc(fpr["micro"], tpr["micro"])

I have some problem with svm and decision_function. Is there a better way to solve this problem?

2

This issue doesn’t seem to be PyTorch-related and I think this scikit-learn tutorial shows the usage for a multi class use case.