aimclub / FEDOT

Automated modeling and machine learning framework FEDOT
https://fedot.readthedocs.io
BSD 3-Clause "New" or "Revised" License
632 stars 86 forks source link

Metric evaluation error: y_true and y_pred contain different number of classes #1016

Closed nicl-nno closed 1 year ago

nicl-nno commented 1 year ago

The problem "Metric evaluation error: y_true and y_pred contain different number of classes 45, 85" raises for click_prediction_small dataset

Looks like class stratification is failing.

curiousHG commented 1 year ago

Can you expand on the issue? Or maybe narrow down to which file or code the problem is?

nicl-nno commented 1 year ago

@curiousHG

This problem occurs for classification tasks when the number of classes is very large and cross-validation fails to preserve the examples for all classes.

The connected problems are for <2 classes in sample and new classes in predict sample.

gkirgizov commented 1 year ago

Duplicate of https://github.com/aimclub/FEDOT/issues/954