[FEA] OneHotEncoder support in sklearn Pipeline

[yasmina-altair]

Is your feature request related to a problem? Please describe.
I would like to use cuml.preprocessing.OneHotEncoder as part of Pipeline so that I can assemble additional steps and multiple preprocessors. However, OneHotEncoder alone in a Pipeline fails. For example:

import cudf
from cuml.preprocessing import OneHotEncoder
from sklearn.pipeline import Pipeline

X = cudf.DataFrame({'a': ["y", "y", "n", "n"]})
cat_transformer = Pipeline(steps=[("onehot", OneHotEncoder())])
cat_transformer.fit_transform(X[["a"]])
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-1-862f3f5c6c73> in <module>
      7     ("onehot", OneHotEncoder())
      8 ])
----> 9 cat_transformer.fit_transform(X)

/opt/conda/envs/rapids/lib/python3.7/site-packages/sklearn/pipeline.py in fit_transform(self, X, y, **fit_params)
    374             fit_params_last_step = fit_params_steps[self.steps[-1][0]]
    375             if hasattr(last_step, 'fit_transform'):
--> 376                 return last_step.fit_transform(Xt, y, **fit_params_last_step)
    377             else:
    378                 return last_step.fit(Xt, y,

TypeError: fit_transform() takes 2 positional arguments but 3 were given

Describe the solution you'd like
OneHotEncoder works with Pipeline similar to sklearn's OneHotEncoder:

import pandas as pd
from sklearn.preprocessing import OneHotEncoder
from sklearn.pipeline import Pipeline

X = pd.DataFrame({'a': ["y", "y", "n", "n"]})
cat_transformer = Pipeline(steps=[("onehot", OneHotEncoder())])
cat_transformer.fit_transform(X)
---------------------------------------------------------------------------
<4x2 sparse matrix of type '<class 'numpy.float64'>'
	with 4 stored elements in Compressed Sparse Row format>

Describe alternatives you've considered
Omit using Pipeline with cuml OneHotEncoder.

Additional context
As a reference, I am using cuml version 0.16.0a+882.g5851f4140 and sklearn version 0.23.1.

@tfeher here's an example with a categorical transformer and sklearn's Pipeline. It would be great to have OneHotEncoder in a Pipeline, I don't see this error with some of the cuml experimental preprocessors.

[viclafargue]

Thank you for opening the issue. It looks like the fit_transform method of the OneHotEncoder class should be able to take a y parameter to be compatible with Scikit-Learn's Pipeline. This will be corrected.

[wphicks]

Looks like this also affects LabelEncoder.

[viclafargue]

Solved with #3192