Reputation: 131
I got this warning
UndefinedMetricWarning: No positive samples in y_true, true positive value should be meaningless UndefinedMetricWarning)
Do you have any idea what it means?
Upvotes: 10
Views: 20739
Reputation: 2686
For me, I got this error when I erroneously passed a pos_label
param.
The docs say the y_true
param should be "binary labels. If labels are not either {-1, 1} or {0, 1}, then pos_label
should be explicitly given."
However my scores happened to be all 0
and 1
omitting the pos_label
parameter solved it for me!
Upvotes: 3
Reputation: 16966
This means that all the values in y_true are zeros, which means there is no positive class records in the given dataset.
Try the following to understand the distribution of classes in your dataset.
from collections import Counter
Counter(y_true) # y_true must be your labels
Upvotes: 7