Hello community!
I am training several models with hyperparam optimisation. I am using
successive halving but I think the question applies to any search algo.
I am optimising for roc_auc_score. Some of the datasets are very imbalanced,
which means that the roc can't be calculated and it is set to 1.
Following error:
UndefinedMetricWarning: No positive samples in y_true, true positive value
should be meaningless
I guess the obvious answer is that I should optimise for something else, like
the balanced accuracy, but the search runs anyways, so I was wondering which
hyperparameters are returned? the first tested?
Thanks a lot!
Best
Sole
_______________________________________________
scikit-learn mailing list -- [email protected]
To unsubscribe send an email to [email protected]
https://mail.python.org/mailman3//lists/scikit-learn.python.org
Member address: [email protected]
