search

dask / dask-searchcv

dask-searchcv is now part of dask-ml: https://github.com/dask/dask-ml
BSD 3-Clause "New" or "Revised" License
240 stars 43 forks source link

Import MaskedArray from numpy #31

Closed mrocklin closed 7 years ago

mrocklin commented 7 years ago

The sklearn import has broken

jcrist commented 7 years ago

What version of scikit-learn are you on? This requires 0.18.1, as there were some bugfixes needed. If you look at the code this should be there since then.

mrocklin commented 7 years ago

I'm on 0.18 (.0?) . This seems to be latest on conda-forge?

mrocklin@carbon:~$ anaconda search scikit-learn | grep conda-forge Using Anaconda Cloud api site https://api.anaconda.org conda-forge/scikit-learn | 0.18 | conda | linux-64, osx-64

Perhaps try-except between sklearn and numpy?

On Thu, Mar 9, 2017 at 5:06 PM, Jim Crist notifications@github.com wrote:

What version of scikit-learn are you on? This requires 0.18.1, as there were some bugfixes needed. If you look at the code https://github.com/scikit-learn/scikit-learn/blob/master/sklearn/utils/fixes.py#L363-L378 this should be there since then.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/dask/dask-learn/pull/31#issuecomment-285498934, or mute the thread https://github.com/notifications/unsubscribe-auth/AASszGSfV5dUbsEdchpTWsSDAru9ti-Pks5rkHf4gaJpZM4MYrAA .

jcrist commented 7 years ago

The fix in scikit-learn is needed to be able to pickle the masked arrays, but that's it, so a try/catch block would probably be fine. You're correct, 0.18.0 is the latest on conda-forge, but not the latest release. Not sure why conda-forge isn't up to date yet.

jcrist commented 7 years ago

This was fixed in #37, it now works with scikit-learn 0.18.0+. Closing.