ModuleNotFoundError: No module named 'sklearn.preprocessing._label'

[greatsharma]

on executing the following code

import os
import joblib
import pandas as pd

INPUT_PATH = os.environ.get('INPUT_PATH')
META_PATH = os.environ.get('META_PATH')
LOGS_PATH = os.environ.get('LOGS_PATH')

all_data_pipe = joblib.load(META_PATH + 'all_data_pipe.pkl')

I am getting the following error at last line joblib.load(META_PATH + 'all_data_pipe.pkl')

Traceback (most recent call last):
  File "/home/winter_soilder/anaconda3/lib/python3.7/runpy.py", line 183, in _run_module_as_main
    mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
  File "/home/winter_soilder/anaconda3/lib/python3.7/runpy.py", line 109, in _get_module_details
    __import__(pkg_name)
  File "/home/winter_soilder/Desktop/MPG/Modelling/src/automl.py", line 16, in <module>
    all_data_pipe = joblib.load(META_PATH + 'all_data_pipe.pkl')
  File "/home/winter_soilder/my_envs/mpg/lib/python3.7/site-packages/joblib/numpy_pickle.py", line 605, in load
    obj = _unpickle(fobj, filename, mmap_mode)
  File "/home/winter_soilder/my_envs/mpg/lib/python3.7/site-packages/joblib/numpy_pickle.py", line 529, in _unpickle
    obj = unpickler.load()
  File "/home/winter_soilder/anaconda3/lib/python3.7/pickle.py", line 1085, in load
    dispatch[key[0]](self)
  File "/home/winter_soilder/anaconda3/lib/python3.7/pickle.py", line 1373, in load_global
    klass = self.find_class(module, name)
  File "/home/winter_soilder/anaconda3/lib/python3.7/pickle.py", line 1423, in find_class
    __import__(module, level=0)
ModuleNotFoundError: No module named 'sklearn.preprocessing._label'

[johannfaouzi]

There is a sklearn/preprocessing/_label.py file in master.

Which version of scikit-learn are you using? Which version of scikit-learn were you using when you save your object with joblib?

[greatsharma]

>>> import sklearn
>>> sklearn.__version__
'0.21.3'
>>> import joblib
>>> joblib.__version__
'0.14.1'

[Smeet97Kathiria]

Did you try to do it with pickle instead of joblib?

[greatsharma]

@Smeet97Kathiria I am working with joblib only. Can you please resolve the issue with joblib itself?

[rth]

Looks like you created this pickle with scikit-learn >=0.22.0 and are trying to load it with scikit-learn <0.22.0.

Pickles are not necessarily compatible across scikit-learn versions so this behavior is expected (and the use case is not supported). For more details, see https://scikit-learn.org/dev/modules/model_persistence.html#model-persistence

[EgoPingvina]

@rth I have scikit-learn 0.24.1, but I still have the same problem

[3-0-8]

@rth I have scikit-learn 0.24.1, but I still have the same problem

me too

[rth]

What version of scikit-learn did you use the create the pickle in question?

[YazmaniReyesH]

Hello good Morning.

I have a similar error but with 'sklearn.preprocessing.data' in PyCharm:

Traceback (most recent call last): File "C:/Users/Yazmani/Desktop/Code/SportPersonClassifier/server/server.py", line 15, in <module> util.load_saved_artifacts() File "C:\Users\Yazmani\Desktop\Code\SportPersonClassifier\server\util.py", line 50, in load_saved_artifacts __model = joblib.load(f) File "C:\Users\Yazmani\AppData\Local\Programs\Python\Python38-32\lib\site-packages\joblib\numpy_pickle.py", line 577, in load obj = _unpickle(fobj) File "C:\Users\Yazmani\AppData\Local\Programs\Python\Python38-32\lib\site-packages\joblib\numpy_pickle.py", line 506, in _unpickle obj = unpickler.load() File "C:\Users\Yazmani\AppData\Local\Programs\Python\Python38-32\lib\pickle.py", line 1212, in load dispatch[key[0]](self) File "C:\Users\Yazmani\AppData\Local\Programs\Python\Python38-32\lib\pickle.py", line 1528, in load_global klass = self.find_class(module, name) File "C:\Users\Yazmani\AppData\Local\Programs\Python\Python38-32\lib\pickle.py", line 1579, in find_class __import__(module, level=0) ModuleNotFoundError: No module named 'sklearn.preprocessing.data'

I am using pickle, and I do not understand why that error appears

I have already updated pip, scikit-learn, etc (I have the latest versions of each package), I have also changed the interpreter between the native python version 3.9 and the Anaconda interpreter also with python 3.9.

Can anybody help me?

[HevOHel]

Hey there, with regardless of other dependencies, sklearn.preprocessing.label is used at or less than 0.21.X, and in contrast, sklearn.preprocessing._label is used as or higher than 0.22.X.
Be careful with the underscode before 'label'.

[amrouchk]

I'm facing the same issue. any solution?

[adrinjalali]

You need to load the pickle files with the same version of sklearn which was use to dump them, otherwise there is not guarantee that the load would work.

[zimonth]

How to know, which version of sklearn was used with some pickle file?
I need to get this one pickle file loaded, but get the same error (without underscore):
ModuleNotFoundError: No module named 'sklearn.preprocessing.label'

Is the sklearn pip package the only one which has to be downgraded, or some other pip packages also?
ls -h
1.1M 2020-06-17 pid736549-resnet512-95_2020-06-17/onehot-pid736549_2020-06-17_12:07:56.101855.pkl

It has been written with:

from sklearn.preprocessing import LabelEncoder
from sklearn.preprocessing import OneHotEncoder
# (lines omitted) 
# onehot is an object of several dictionaries where some of them have LaberEncoder objects.
import pickle
with open(filename, 'wb') as f:
    pickle.dump(onehot, f, pickle.HIGHEST_PROTOCOL) 

When trying to use this:

onehot = pickle.load(open(load_onehot_file, "rb"))

I get this:

    onehot = pickle.load(open(load_onehot_file, "rb"))
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
**ModuleNotFoundError: No module named 'sklearn.preprocessing.label'**

If I now use fresh start to pickle.dump and picle.load, it works fine. But I'd need to load that old pickle file.

Scikit-learn (sklearn) version:

# python3 -m pip list|grep scikit-learn
scikit-learn             1.3.0

>>> import sklearn
>>> sklearn.__version__
'1.3.0'