How to use auc roc curve for multi-class model in fast ai

My Dataset Looks Like this a csv file and a bunch of images in a folder :

I want calculate auc_roc_score for each class as my prediction output.
How to use auc_roc_score as metric to get a prediction like this :slight_smile:

@jeremy Please help in answering this .

Here is how to do it:

from sklearn.metrics import roc_auc_score
def comp_metric(preds, targs, labels=range(len(LABEL_COLS))):
    # One-hot encode targets
    targs = np.eye(4)[targs]
    return np.mean([roc_auc_score(targs[:,i], preds[:,i]) for i in labels])

def healthy_roc_auc(*args):
    return comp_metric(*args, labels=[0])

def multiple_diseases_roc_auc(*args):
    return comp_metric(*args, labels=[1])

def rust_roc_auc(*args):
    return comp_metric(*args, labels=[2])

def scab_roc_auc(*args):
    return comp_metric(*args, labels=[3])
            AccumMetric(healthy_roc_auc, flatten=False),
            AccumMetric(multiple_diseases_roc_auc, flatten=False),
            AccumMetric(rust_roc_auc, flatten=False),
            AccumMetric(scab_roc_auc, flatten=False),
            AccumMetric(comp_metric, flatten=False)

learn = cnn_learner(dls, resnet34,metrics=[accuracy,*metrics_])