noccn
noccn copied to clipboard
AUC value
Hello Daniel,
It appears that 'auc-value' (in the predict.py, def report) is only used for binary classification. Given multiple classification problem, it's likely to throw out error. I wonder if some updates can be done to fixed, e.g. check the number of classes (len(set(y_true)), and only computer auc if it's a binary classification problem.
Thanks, Vanessa