ImportError: DLL load failed when importing sklearn in Jupyter with Anaconda

Question

I updated Anaconda, and since then I can't import sklearn in my Jupyter Notebook. Here is my traceback:

---------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
<ipython-input-2-d20256e4506d> in <module>
      4 
      5 # Scikit-Learn ≥0.20 is required
----> 6 import sklearn
      7 # assert sklearn.__version__ >= "0.20"
      8 

~\Anaconda3\lib\site-packages\sklearn\__init__.py in <module>
     80     from . import _distributor_init  # noqa: F401
     81     from . import __check_build  # noqa: F401
---> 82     from .base import clone
     83     from .utils._show_versions import show_versions
     84 

~\Anaconda3\lib\site-packages\sklearn\base.py in <module>
     18 
     19 from . import __version__
---> 20 from .utils import _IS_32BIT
     21 
     22 _DEFAULT_TAGS = {

~\Anaconda3\lib\site-packages\sklearn\utils\__init__.py in <module>
     25 from ..exceptions import DataConversionWarning
     26 from .deprecation import deprecated
---> 27 from .fixes import np_version
     28 from .validation import (as_float_array,
     29                          assert_all_finite,

~\Anaconda3\lib\site-packages\sklearn\utils\fixes.py in <module>
     16 import scipy.sparse as sp
     17 import scipy
---> 18 import scipy.stats
     19 from scipy.sparse.linalg import lsqr as sparse_lsqr  # noqa
     20 

~\Anaconda3\lib\site-packages\scipy\stats\__init__.py in <module>
    382 from __future__ import division, print_function, absolute_import
    383 
--> 384 from .stats import *
    385 from .distributions import *
    386 from .morestats import *

~\Anaconda3\lib\site-packages\scipy\stats\stats.py in <module>
    183 import scipy.special as special
    184 from scipy import linalg
--> 185 from . import distributions
    186 from . import mstats_basic
    187 from ._stats_mstats_common import (_find_repeats, linregress, theilslopes,

~\Anaconda3\lib\site-packages\scipy\stats\distributions.py in <module>
      8 from __future__ import division, print_function, absolute_import
      9 
---> 10 from ._distn_infrastructure import (entropy, rv_discrete, rv_continuous,
     11                                     rv_frozen)
     12 

~\Anaconda3\lib\site-packages\scipy\stats\_distn_infrastructure.py in <module>
     23 
     24 # for root finding for discrete distribution ppf, and max likelihood estimation
---> 25 from scipy import optimize
     26 
     27 # for functions of continuous distributions (e.g. moments, entropy, cdf)

~\Anaconda3\lib\site-packages\scipy\optimize\__init__.py in <module>
    388 
    389 from .optimize import *
--> 390 from ._minimize import *
    391 from ._root import *
    392 from ._root_scalar import *

~\Anaconda3\lib\site-packages\scipy\optimize\_minimize.py in <module>
     28 from ._trustregion_krylov import _minimize_trust_krylov
     29 from ._trustregion_exact import _minimize_trustregion_exact
---> 30 from ._trustregion_constr import _minimize_trustregion_constr
     31 
     32 # constrained minimization

~\Anaconda3\lib\site-packages\scipy\optimize\_trustregion_constr\__init__.py in <module>
      2 
      3 
----> 4 from .minimize_trustregion_constr import _minimize_trustregion_constr
      5 
      6 __all__ = ['_minimize_trustregion_constr']

~\Anaconda3\lib\site-packages\scipy\optimize\_trustregion_constr\minimize_trustregion_constr.py in <module>
      2 import time
      3 import numpy as np
----> 4 from scipy.sparse.linalg import LinearOperator
      5 from .._differentiable_functions import VectorFunction
      6 from .._constraints import (

~\Anaconda3\lib\site-packages\scipy\sparse\linalg\__init__.py in <module>
    114 from .dsolve import *
    115 from .interface import *
--> 116 from .eigen import *
    117 from .matfuncs import *
    118 from ._onenormest import *

~\Anaconda3\lib\site-packages\scipy\sparse\linalg\eigen\__init__.py in <module>
      9 from __future__ import division, print_function, absolute_import
     10 
---> 11 from .arpack import *
     12 from .lobpcg import *
     13 

~\Anaconda3\lib\site-packages\scipy\sparse\linalg\eigen\arpack\__init__.py in <module>
     20 from __future__ import division, print_function, absolute_import
     21 
---> 22 from .arpack import *

~\Anaconda3\lib\site-packages\scipy\sparse\linalg\eigen\arpack\arpack.py in <module>
     43 __all__ = ['eigs', 'eigsh', 'svds', 'ArpackError', 'ArpackNoConvergence']
     44 
---> 45 from . import _arpack
     46 import numpy as np
     47 import warnings

ImportError: DLL load failed: La procédure spécifiée est introuvable.

I have found several threads about this problem but none of them works for me. Apparently there's a problem with arpack in my case. I already tried to uninstal/reinstall numpy, scipy, sklearn, mkl. Nothing is working. Please help

Answer 1

I've finally found a solution. I reinstalled numpy+mkl. Apparently you have to download it specifically from here.

Then in the Terminal you go to the folder you saved the file and you run the following command:

pip install --upgrade --force-reinstall "numpy-1.18.1+mkl-cp37-cp37m-win_amd64.whl"

(replace the name of the file by the one you have downloaded)

Magic. It's working.