A bug in the partial_fit function

[yousefabdi]

Hello,

At first, I would like to thank for developing this great package. It was nececcary for stream mining research community.

I have a problem with "partial_fit" function when using on multi-class datasets. The error message I get is as follows: ValueError: could not broadcast input array from shape (2,) into shape (3,)

I attached the dataset. RBF_Gradual_3Class_.csv

When I set the "classes" argument of the "partial_fit" in the KUE algorithm, it works. clf.partial_fit(X_train, y_train, classes=labels)

However, in other algorithms, UOB for example, I get the same error by setting the "classes" argument too.

❌❌❌ The error is on self.current_tdcs_

[yousefabdi]

Most of the multi-class algorithms have issues with multi-class datasets.

❌ NIE.py: on f1-score