ImportError: No module named sklearn.cross_validation

Question

I am using python 2.7 in Ubuntu 14.04. I installed scikit-learn, numpy and matplotlib with these commands:

sudo apt-get install build-essential python-dev python-numpy \
python-numpy-dev python-scipy libatlas-dev g++ python-matplotlib \
ipython

But when I import these packages:

from sklearn.cross_validation import train_test_split

It returns me this error:

ImportError: No module named sklearn.cross_validation

What I need to do?

perhaps your module name is wrong if you have installed sklearn and anaconda correctly. — CKM, Oct 11 '16 at 09:52
I am really just repeating it, but you have to use sklearn.model_selection from now on. cross_validation is not usable since sklearn 20+ — Michal Mikuláši, Mar 23 '19 at 18:51
Wow 13 answers to say the same thing. The latest 4 years after the first one (in case bits would fade with time I suppose) — mins, Jan 09 '21 at 18:12

score 804 · Accepted Answer · edited Nov 26 '19 at 14:14

804

It must relate to the renaming and deprecation of cross_validation sub-module to model_selection. Try substituting cross_validation to model_selection

edited Nov 26 '19 at 14:14

The Bearded Llama

3,036
4
20
31

answered Jan 17 '16 at 22:09

Dima Lituiev

12,544
10
41
58

8

Why didn't sklearn developers put in an alias for backward compatibility? Also, the doc for that older version should indicate this refactor: https://scikit-learn.org/0.16/modules/generated/sklearn.cross_validation.train_test_split.html. – flow2k Aug 15 '19 at 22:23
1

Because making code inoperable with API changes is one of the favorite features of python-based data science ;-) (See Pandas decade of version 0.x changes) – Jeff Winchell Nov 12 '21 at 14:52

score 186 · Answer 2 · edited Nov 22 '18 at 03:03

186

train_test_split is now in model_selection. Just type:

from sklearn.model_selection import train_test_split

it should work

edited Nov 22 '18 at 03:03

Jee Mok

6,157
8
47
80

answered Nov 22 '18 at 01:51

ayat ullah sony

1,973
1
10
7

score 45 · Answer 3 · edited Apr 15 '19 at 10:04

45

I guess cross selection is not active anymore. We should use instead model selection. You can write it to run, from sklearn.model_selection import train_test_split

Thats it.

edited Apr 15 '19 at 10:04

Jeru Luke

20,118
13
80
87

answered Mar 24 '19 at 09:32

MSAHIN

451
4
4

This is the Answer. You are awesome. – Biswajit Panday Apr 24 '19 at 20:55

score 36 · Answer 4 · edited Dec 30 '19 at 21:40

36

Make sure you have Anaconda installed and then create a virtualenv using conda. This will ensure all the imports work

Python 2.7.9 |Anaconda 2.2.0 (64-bit)| (default, Mar  9 2015, 16:20:48) 
[GCC 4.4.7 20120313 (Red Hat 4.4.7-1)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
Anaconda is brought to you by Continuum Analytics.
Please check out: http://continuum.io/thanks and https://binstar.org
>>> from sklearn.cross_validation import train_test_split

edited Dec 30 '19 at 21:40

Cody Gray - on strike

239,200
50
490
574

answered Jun 05 '15 at 13:19

Richard Rublev

7,718
16
77
121

No, I don't have it, I need to install Anaconda? – arthurckl Jun 05 '15 at 13:21
Well,I would suggest you to install it.But You can also try sudo apt-get install python-sklearn,that should work. – Richard Rublev Jun 05 '15 at 13:30
7

The answer by @DimaLituiev is really the correct answer here. This issue is mostly related to a renaming, not a true missing package. – Amrinder Arora Feb 08 '19 at 21:47
3

This isn't necessarily needed. You are suggesting installing a new environment just for one package that was renamed, I would say this solution is an overkill. Dima's solution should be marked as correct answer – MuhsinFatih Mar 21 '19 at 14:35
I guess the question should have addressed if he has "import sklearn" worked or not first. If it works, maybe this is not a naming issue. – ju. Apr 01 '19 at 17:30
Note that you don't have to have Anaconda to fix this issue (I just use pip). See other answers for a possible fix. – Rushi Agrawal May 17 '19 at 00:31

score 30 · Answer 5 · edited Jul 26 '19 at 16:42

30

sklearn.cross_validation is now changed to sklearn.model_selection

Just use

from sklearn.model_selection import train_test_split

I think that will work.

edited Jul 26 '19 at 16:42

akshayk07

2,092
1
20
32

answered Jul 26 '19 at 14:27

Tanmoy Bhowmick

1,305
15
20

score 24 · Answer 6 · answered Jun 30 '19 at 20:32

24

sklearn.cross_validation

has changed to

sklearn.model_selection

Checkout the documentation here: https://scikit-learn.org/stable/modules/cross_validation.html

answered Jun 30 '19 at 20:32

Asad

930
7
10

nantitv · Answer 7 · 2018-11-04T14:36:14.810

17

May be it's due to the deprecation of sklearn.cross_validation. Please replace sklearn.cross_validation with sklearn.model_selection

Ref- https://github.com/amueller/scipy_2015_sklearn_tutorial/issues/60

edited Nov 04 '18 at 14:36

answered Nov 01 '18 at 15:21

nantitv

3,539
4
38
61

2

It appears your answer is the [same as a previous given](https://stackoverflow.com/a/34844352/10400050) in the same thread. If you think that answer can be improved, consider [editing](https://stackoverflow.com/posts/34844352/edit) it instead of making a new answer to the same question. – Johan Nov 04 '18 at 14:40
@Johan Some how I did not see the other answer. I could delete my answer if you prefer. – nantitv Nov 04 '18 at 23:15

score 13 · Answer 8 · edited Feb 10 '19 at 18:58

13

Splitting the dataset into the Training set and Test set

from sklearn.model_selection import train_test_split

edited Feb 10 '19 at 18:58

Eric Aya

69,473
35
181
253

answered Feb 10 '19 at 17:51

optimists

191
1
7

2

Same solution as ayat ullah sony's answer. – Eric Aya Feb 10 '19 at 18:58

score 10 · Answer 9 · answered Mar 27 '19 at 16:00

Past : from sklearn.cross_validation (This package is deprecated in 0.18 version from 0.20 onwards it is changed to from sklearn import model_selection).

Present: from sklearn import model_selection

Example 2:

Past : from sklearn.cross_validation import cross_val_score (Version 0.18 which is deprecated)

Present : from sklearn.model_selection import cross_val_score

score 10 · Answer 10 · answered Oct 09 '19 at 09:55

10

sklearn.cross_validation is now changed to sklearn.model_selection

Just change

sklearn.cross_validation

to

sklearn.model_selection

answered Oct 09 '19 at 09:55

Sani Kamal

1,208
16
26

score 10 · Answer 11 · edited Apr 19 '21 at 07:57

10

The cross_validation is not available anymore.

Try to use model_selection instead of cross_validation:

from sklearn.model_selection import train_test_split

edited Apr 19 '21 at 07:57

Ewran

328
4
15

answered Apr 19 '21 at 06:05

Abhishek

121
1
4

jss367 · Answer 12 · 2020-04-01T16:45:46.520

If you have code that needs to run various versions you could do something like this:

import sklearn
if sklearn.__version__ > '0.18':
    from sklearn.model_selection import train_test_split
else:
    from sklearn.cross_validation import train_test_split

This isn't ideal though because you're comparing package versions as strings, which usually works but doesn't always. If you're willing to install packaging, this is a much better approach:

from packaging.version import parse
import sklearn
if parse(sklearn.__version__) > parse('0.18'):
    from sklearn.model_selection import train_test_split
else:
    from sklearn.cross_validation import train_test_split

score 5 · Answer 13 · answered Dec 04 '19 at 05:22

5

change the code like this

# from sklearn.cross_validation import train_test_split
from sklearn.model_selection import train_test_split

answered Dec 04 '19 at 05:22

Muhammad Shabin

381
5
5

score 4 · Answer 14 · edited May 01 '20 at 10:15

4

train_test_split is part of the module sklearn.model_selection, hence, you may need to import the module from model_selection

Code:

from sklearn.model_selection import train_test_split

edited May 01 '20 at 10:15

David Buck

3,752
35
31
35

answered May 01 '20 at 08:51

Noah Sheldon

1,461
10
12

score 3 · Answer 15 · answered Sep 19 '19 at 10:06

3

cross_validation was deprecated some time ago, try switching it out with model_selection

answered Sep 19 '19 at 10:06

Aniket Mukherjee

45
3

try adding more explanation to your answer to give it more value. – Zeina Sep 19 '19 at 10:29

ImportError: No module named sklearn.cross_validation

15 Answers15

Linked

Related