Patch which makes the module work in Python 3

[amacbee]

Here is a patch which makes the module work in Python 3..
isinstance(basestring) raises a NameError in Python 3. because type of basestring is not supported.

[xgdgsc]

Any dev looking at this? I also get the error NameError: name 'basestring' is not defined using python3

[ThomasP6t]

This can be more easily solved by modifying setup.py (see my pull request); upon installation the correct version of the condition will be in the code.

[ghost]

I tried to apply the proposed debug, but i'm ending up, getting the error:
TypeError: fit() missing 1 required positional argument: 'X'
in sklearn_pandas/init.py", line 134, in fit
(raised by sklearn/base.py", line 426, in fit_transform)
is there any other fix i can apply?
Edit: I'm a dummy who did not initialize the sklearn.preprocessing class.

[ThomasP6t]

MS-Position, can you give us a working example of the code that generates this error?

[ghost]

I actually can not (see above). The patch does indeed work.
Thank You for Your post.

[dukebody]

I think this PR is not needed anymore since the Py3K problem was resolved with #15

[calpaterson]

I'd like to find a better way to do it than mutating basestring, but it does seem that this problem is solved at least for now

[dukebody]

We can look into six: https://pypi.python.org/pypi/six

[ThomasP6t]

I made a new pull request which implements this using six: #32

[ThomasP6t]

And another that does the same thing without the need for six: #33