python - ModuleNotFoundError: No module named 'numpy.testing.nosetester'

36
votes

I was using the Decision Tree and this error was raised. The same situation appeared when I used Back Propagation. How can I solve it?

import pandas as pd
import numpy as np
a = np.test()
f = open('E:/lgdata.csv')
data = pd.read_csv(f,index_col = 'id')

x = data.iloc[:,10:12].as_matrix().astype(int)
y = data.iloc[:,9].as_matrix().astype(int)

from sklearn.tree import DecisionTreeClassifier as DTC
dtc = DTC(criterion='entropy')
dtc.fit(x,y)
x=pd.DataFrame(x) 

from sklearn.tree import export_graphviz
with open('tree.dot','w') as f1:
    f1 = export_graphviz(dtc, feature_names = x.columns, out_file = f1)

Traceback (most recent call last):
  File "<ipython-input-40-4359c06ae1f0>", line 1, in <module>
    runfile('C:/ProgramData/Anaconda3/lib/site-packages/scipy/_lib/_numpy_compat.py', wdir='C:/ProgramData/Anaconda3/lib/site-packages/scipy/_lib')
  File "C:\ProgramData\Anaconda3\lib\site-packages\spyder\utils\site\sitecustomize.py", line 710, in runfile
    execfile(filename, namespace)
  File "C:\ProgramData\Anaconda3\lib\site-packages\spyder\utils\site\sitecustomize.py", line 101, in execfile
    exec(compile(f.read(), filename, 'exec'), namespace)
  File "C:/ProgramData/Anaconda3/lib/site-packages/scipy/_lib/_numpy_compat.py", line 9, in <module>
    from numpy.testing.nosetester import import_nose

ModuleNotFoundError: No module named 'numpy.testing.nosetester'

8
pythonnumpymachine-learningimporterrornose

8 Answers

56
votes

This is happening due to a version incompatibility between numpy and scipy. numpy in its latest versions have deprecated numpy.testing.nosetester.

Replicating the issue

pip install numpy==1.18 # > 1.18
pip install scipy<=0.19.0 # <= 0.19

and

from sklearn.tree import DecisionTreeClassifier as DTC

Triggers the error.

Fixing the error

Upgrade your scipy to a higher version.

pip install numpy==1.18
pip install scipy==1.1.0
pip install scikit-learn==0.21.3

But not limited to this. By upgrading the above libraries to the latest stable, you should be able to get rid of this error.

6
votes

I needed to upgrade scipy

pip3 install -U scipy

4
votes

I was facing the same error while using lexnlp package Got it fixed by installing:

scipy==1.4.1
pandas==0.23.4    
numpy==1.18.1
lexnlp==0.2.7.1

(Only install lexnlp if know you're explicitly using it in your project and you know what you're doing)

4
votes

I also faced same issue while loading the model and fixed by upgrading below libraries

pip install -U numpy
pip install -U pandas
pip install -U scipy
3
votes

I solved this by:

pip uninstall numpy
pip install numpy == 1.17.0

and using:

from numpy.testing import rundocs
2
votes

try installing numpy version 1.17.0 using pip or pip3 (assuming you already installed pip3)

pip3 install numpy==1.17.0
0
votes

If you are using Jetson TX2 or any other aarch64 based device. You can solve the issue by installing latest numpy and scipy libraries.

This also works for x86 based systems. (You can skip libatlas-base-dev and gfortran for x86 based systems)

sudo pip3 install numpy==1.19.0
sudo apt-get install libatlas-base-dev gfortran
sudo pip3 install scipy==1.5.1
0
votes

For me it solved by this link, apparently an open issue. Downgrade to numpy==1.16.4