Closed
Description
While testing the class_weight parameter in several estimators to investigate #1411, I noticed that the parameter doesn't work as I would expect in RidgeClassifier either.
Maybe this is some misunderstanding on my part or some regularization issue, but if I have noisy labels, I would have expected to be be able to move the decision boundary.
This is not the case as illustrated in this notebook
Any input would be appreciated.