You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Wendy Wong commented: This is not so simple. When a user provides a training dataset with weights, but provides a test/validation dataset without weights, they should receive a warning that they are not getting weighted metrics and all rows are considered equal. If this is what they want, they can just ignore the warning.
This warning gets to be incorrect when there is cross validation and a validation dataset is provided. In this case, the weight column is internally generated by H2O-3 to divide up the training datasets into folds. This has nothing to do with weighted training. In this case, no warning should be generated.
No description provided.
The text was updated successfully, but these errors were encountered: