edit _transform_count_encode

[pyrrhull]

moved X.fillna... inside for loop so that it handles only columns
present in self.cols individually and not the whole dataframe
addition to #260

[janmotl]

Looks good to me. Consider adding a unit test to make sure that the issue does not return (personally, I would have preffered a parametrized test in test_encoders to cover all the encoders).

[pyrrhull]

I am considering the tests but don't know how to proceed. What should the tests be against and which parameters to use? Isn't test_metamorphic already doing the job here?

[janmotl]

Metamorphic tests encoded features. But we also have to test non-encoded features. The test to add:

    def test_ignored_columns_are_untouched(self):
        # Make sure None values in ignored columns are preserved.
        # See: https://github.com/scikit-learn-contrib/category_encoders/pull/261
        X = pd.DataFrame({'col1': ['A', 'B', None], 'col2': ['C', 'D', None]})
        y = [1, 0, 1]

        for encoder_name in (set(encoders.__all__)):
            with self.subTest(encoder_name=encoder_name):
                enc = getattr(encoders, encoder_name)(cols=['col1'])
                out = enc.fit_transform(X, y)
                self.assertTrue(out.col2[2] is None)

[pyrrhull]

Great, so I added your test to test_encoders.py.

[wdm0006]

Working through some old PRs, this looks good. If you'd be interested in working to maintain the project more actively please check out #248, I'm (as you probably noticed) not able to consistently.