Cannot import sklearn.model_selection in scikit-learn

Question

I am trying to import sklearn.model_selection. I have tried to reinstall scikit-learn and anaconda, still not working. Here is the error msg I got:

ImportError                               Traceback (most recent call last)
<ipython-input-69-e49df3a70ea4> in <module>()
      4 get_ipython().magic(u'matplotlib inline')
      5 # from sklearn.model_selection import train_test_split
----> 6 import sklearn.model_selection

/Users/Lu/anaconda/lib/python2.7/site-packages/sklearn/model_selection/__init__.py in <module>()
----> 1 from ._split import BaseCrossValidator
      2 from ._split import KFold
      3 from ._split import GroupKFold
      4 from ._split import StratifiedKFold
      5 from ._split import TimeSeriesSplit

/Users/Lu/anaconda/lib/python2.7/site-packages/sklearn/model_selection/_split.py in <module>()
     34 from ..utils.random import choice
     35 from ..base import _pprint
---> 36 from ..gaussian_process.kernels import Kernel as GPKernel
     37 
     38 __all__ = ['BaseCrossValidator',

/Users/Lu/anaconda/lib/python2.7/site-packages/sklearn/gaussian_process/__init__.py in <module>()
     11 """
     12 
---> 13 from .gpr import GaussianProcessRegressor
     14 from .gpc import GaussianProcessClassifier
     15 from . import kernels

/Users/Lu/anaconda/lib/python2.7/site-packages/sklearn/gaussian_process/gpr.py in <module>()
     10 import numpy as np
     11 from scipy.linalg import cholesky, cho_solve, solve_triangular
---> 12 from scipy.optimize import fmin_l_bfgs_b
     13 
     14 from sklearn.base import BaseEstimator, RegressorMixin, clone

/Users/Lu/anaconda/lib/python2.7/site-packages/scipy/optimize/__init__.py in <module>()
    232 from .optimize import *
    233 from ._minimize import *
--> 234 from ._root import *
    235 from .minpack import *
    236 from .zeros import *

/Users/Lu/anaconda/lib/python2.7/site-packages/scipy/optimize/_root.py in <module>()
     17 
     18 from .optimize import MemoizeJac, OptimizeResult, _check_unknown_options
---> 19 from .minpack import _root_hybr, leastsq
     20 from ._spectral import _root_df_sane
     21 from . import nonlin

/Users/Lu/anaconda/lib/python2.7/site-packages/scipy/optimize/minpack.py in <module>()
      2 
      3 import warnings
----> 4 from . import _minpack
      5 
      6 import numpy as np

ImportError: cannot import name _minpack

Answer 1

Check your scikit-learn version;

import sklearn
print(sklearn.__version__)

sklearn.model_selection is available for version 0.18.1.

What you need to import depends on what you require. For instance, in version 0.18.1, GridSearchCV can be imported as

from sklearn.model_selection import GridSearchCV

Whereas in version 0.17.1, the same can be imported as

from sklearn.grid_search import GridSearchCV

If you find anything in the new scikit documentation that doesn't work for you in your system, then search the document for the current version you are using. The import path might be different but the overall functionality ought to be the same.

If you don't have any previous projects or code that requires the older version, the better option would be to update your scikit-learn package. As you stated that you use Anaconda, the following post would be useful:

How to upgrade scikit-learn package in anaconda