ImportError: cannot import name 'Imputer' from 'sklearn.preprocessing'

[ZoranPandovski]

Describe the bug Installing the latest version of lightwood throws ImportError. I guess the issue is related to the sciki-learn.

To Reproduce Steps to reproduce the behavior:

1. Train model, the issue is not related to a specific dataset
2. See error

from cesium import featurize

File "/home/zoran/MyProjects/lightwood/l/lib/python3.7/site-packages/cesium-0.9.9-py3.7-linux-x86_64.egg/cesium/featurize.py", line 10, in from sklearn.preprocessing import Imputer ImportError: cannot import name 'Imputer' from 'sklearn.preprocessing' (/home/zoran/MyProjects/lightwood/l/lib/python3.7/site-packages/scikit_learn-0.22rc3-py3.7-linux-x86_64.egg/sklearn/preprocessing/init.py)

Screenshots

[ZoranPandovski]

This is related with pip installing latest verzion of scikit-learn which was pre-release 0.22rc3 version. We should use fixed version that works 0.21.3. I will create PR

[ZoranPandovski]

A new scikit-learn version 0.22 was released yesterday, but there are other errors with using that version. We should keep on 0.21.3 for now.

[ZoranPandovski]

Fixed by #74

[AnasK95]

from sklearn.impute import SimpleImputer will work because of the following

DeprecationWarning: Class Imputer is deprecated; Imputer was deprecated in version 0.20 and will be removed in 0.22. Import impute.SimpleImputer from sklearn instead.

[subhashi]

Thank you @AnasK95 It worked.

So I had to use,

from sklearn.impute import SimpleImputer 
imputer = SimpleImputer(missing_values=np.nan, strategy='mean')

instead of

from sklearn.preprocessing import Imputer
imputer = Imputer(missing_values='NaN', strategy='mean', axis=0)

Otherwise, it gives the following error.

from sklearn.preprocessing import Imputer
ImportError: cannot import name 'Imputer' from 'sklearn.preprocessing'

[KavishGoyal]

Thanks @AnasK95 and @subhashi It worked for me as well.

[AnujJha-stack]

thanks it works

[jspw]

but 'axis' is no more a argument there ! btw tahnks @AnasK95 it worked!

[harmeshm]

Thank you @AnasK95.

[Rawkush]

Thank you @AnasK95 It worked. but how do we use axis argument functionality now?

[MutumaIan]

Thanks a lot.

[diegoecon-ai]

Sincerely thanks! @AnasK95

[Norni]

Thanks @subhashi

[ahmedtamam721]

i try to upgrade it but !!!

[dpk-a7]

hey, In version 0.22.2, for error: ImportError: cannot import name 'Imputer' from 'sklearn.preprocessing' (C:\Users\Anaconda3\lib\site-packages\sklearn\preprocessing__init__.py)

You can use this snippet, and it will work fine: (C:\Users\Anaconda3\lib\site-packages\sklearn\preprocessing__init__.py)

from .imputation import Imputer

all = [ . . .

'Imputer',             #  type 'imputer'

. . . ] <-------------------------------------------------------------> Hope this would help <------------------------------------------------------------> full init.py code here:

""" The :mod:sklearn.preprocessing module includes scaling, centering, normalization, binarization and imputation methods. """

from ._function_transformer import FunctionTransformer

from .data import Binarizer from .data import KernelCenterer from .data import MinMaxScaler from .data import MaxAbsScaler from .data import Normalizer from .data import RobustScaler from .data import StandardScaler from .data import QuantileTransformer from .data import add_dummy_feature from .data import binarize from .data import normalize from .data import scale from .data import robust_scale from .data import maxabs_scale from .data import minmax_scale from .data import quantile_transform from .data import power_transform from .data import PowerTransformer from .data import PolynomialFeatures

from ._encoders import OneHotEncoder from ._encoders import OrdinalEncoder

from .label import label_binarize from .label import LabelBinarizer from .label import LabelEncoder from .label import MultiLabelBinarizer

from ._discretization import KBinsDiscretizer

from .imputation import Imputer

all = [ 'Binarizer', 'FunctionTransformer', 'Imputer', 'KBinsDiscretizer', 'KernelCenterer', 'LabelBinarizer', 'LabelEncoder', 'MultiLabelBinarizer', 'MinMaxScaler', 'MaxAbsScaler', 'QuantileTransformer', 'Normalizer', 'OneHotEncoder', 'OrdinalEncoder', 'PowerTransformer', 'RobustScaler', 'StandardScaler', 'add_dummy_feature', 'PolynomialFeatures', 'binarize', 'normalize', 'scale', 'robust_scale', 'maxabs_scale', 'minmax_scale', 'label_binarize', 'quantile_transform', 'power_transform', 'Imputer' ]

[Yassir-Alharbi]

This was sorted by, https://github.com/scikit-learn/scikit-learn/issues/16152

[jhhalls]

This is related with pip installing latest verzion of scikit-learn which was pre-release 0.22rc3 version. We should use fixed version that works 0.21.3. I will create PR

How do i downgrade scikit-learn from 0.22.2 to 0.21

[Etherum7]

Thanks @AnasK95 and @subhashi it helps Cheers to all who are in hadelin machine learning course

[ambujbhargava]

For me I used SimpleImputer

from sklearn.impute import SimpleImputer

worked.

[rohitbajpai1]

from sklearn.impute import SimpleImputer

imputer = SimpleImputer(missing_values = np.nan, strategy='mean') X.fit[:, 1:3] = imputer.fit_transform(X[:, 1:3]) X = X.apply(lambda x: x.fillna(x.value_counts().index[0]))

(slice(None, None, None), slice(1, 3, None))' is an invalid key

Still getting this error

[sudhirp93]

ImportError: cannot import name 'MinMaxScalar' from sklearn.preprocessing .... any help?

[PatChandresh]

I used from sklearn.impute import SimpleImputer imputer = SimpleImputer(strategy='median')

instead of from sklearn.preprocessing import Imputer imputer = Imputer(strategy="median")

not throwing any error

[arefehsajedi]

Hey there,

I just confronted similar error, and I found out the reason was version of the sciki-learn.. in the latest version , you should code as : from sklearn.impute import SimpleImputer instead of : from sklearn.preprocessing import Imputer

also note that inputs are as following: SimpleImputer(missing_values=np.nan, strategy='mean')

also regarding using np, you should summon numpy library. besides, there is no required parameter as axis=0 or 1