Running Jupyter notebook in a virtualenv: installed sklearn module not available

Question

I have installed a created a virtualenv machinelearn and installed a few python modules (pandas, scipy and sklearn) in that environment.

When I run jupyter notebook, I can import pandas and scipy in my notebooks - however, when I try to import sklearn, I get the following error message:

import sklearn

---------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
<ipython-input-1-8fd979e02004> in <module>()
----> 1 import sklearn

ImportError: No module named 'sklearn'

I am able to import all modules, at the command line - so I know they have been successfully installed:

(machinelearn) me@yourbox:~/path/to/machinelearn$ python -c "import pandas, scipy, sklearn"
(machinelearn) me@yourbox:~/path/to/machinelearn$ 

How can I import sklearn in my jupyter notebook running in a virtualenv?

Solution 1

You probably have not installed jupyter / IPython in your virtualenv. Try the following:

python -c "import IPython"

and check that the jupyter command found in your $PATH is the one from the bin folder of your venv:

which jupyter

For windows users in a powershell console, you can use the following to check that the jupyter command in your $env:Path is the one from the Scripts folder of you venv:

get-command jupyter

Edit: if this is the problem, just run python -m pip install jupyter in your venv.

Edit 2: actually you might also need:

python -m ipykernel install --user --name=my-virtualenv-name

and then switch the kernel named "my-virtualenv-name" in the jupyter user interface.

Edit 3: maybe the --user flag in the last command is a bad idea:

python -m ipykernel install --name=my-virtualenv-name

Solution 2

Another approach to take is to have one global jupyter installation, but to point to different kernels to run as the backend.

That approach is outlined here in their docs: http://help.pythonanywhere.com/pages/IPythonNotebookVirtualenvs

Copying below in case the link breaks: You can use a virtualenv for your IPython notebook. Follow the following steps:

Install the ipython kernel module into your virtualenv

workon my-virtualenv-name  # activate your virtualenv, if you haven't already
pip install ipykernel

Now run the kernel "self-install" script:

python -m ipykernel install --user --name=my-virtualenv-name

Replacing the --name parameter as appropriate.

You should now be able to see your kernel in the IPython notebook menu: Kernel -> Change kernel and be able so switch to it (you may need to refresh the page before it appears in the list). IPython will remember which kernel to use for that notebook from then on.

Solution 3

To use Jupyter notebook with virtual environment (using virtualenvwrapper) plus packages installed in that environment, follow steps below:

1. create a virtual environment

   mkvirtualenv --no-site-packages --python=/your/python/path your_env_name
   

2. Activate the virtual environment

   workon your_env_name
   

3. Install Jupyter and other packages

   pip install jupyter, numpy
   

4. Add a new kernel to your Jupyter config

   ipython kernel install --user --name=your_env_name
   

5. Done. You may now use Jupyter notebook under the virtual environment.

   jupyter-notebook
   

Disclaimer: the question has been answered but is hidden in one of the replies. I googled and took sometime to find the right answer. So I just summarize it so someone having the same issue can easily follow.

Solution 4

Assuming that jupyter is installed on your machine, not on the virtual environtment.

Using a virtual environment with Jupyter notebook

VENV_NAME = "YOUR VIRTUAL ENV NAME"

1) virtualenv VENV_NAME

2) source venv/bin/activate

3) Add this package if not present: pip3 install ipykernel

4) Then execute this command: ipython kernel install --user --name=VENV_NAME

5) Now open up the Jupyter Notebook and in change kernel select VENV_NAME

6) To install a new package perform pip3 install <PACKAGE NAME> in your terminal and repeat step 4.

Hope it helps!

Solution 5

Solution without adding a new kernel globally!!

1. create a new virtual environment by

python3 -m virtualenv envname

2. Activate your enviroment and install jupyter in it by

pip install jupyter

One thing you have to make sure before installing jupyter is that you don't have following packages already installed in it.

ipykernel             
ipython               
ipython-genutils      
ipywidgets            
jupyter               
jupyter-client        
jupyter-console     
jupyter-core 

If you've previously installed them then first uninstall them by pip uninstall.

3. Install your desired packages in activated virtualenv and launch jupyter in it and voila!

View more solutions

Comments

[joydeep bhattacharjee]

I think you have not started your notebook with the correct python executable. Its written towards the right of the page and there should be a dropdown. Check if it is the conda installable with the virtualenv and not root.

[Homunculus Reticulli]

Interesting, I don't have IPython installed in the virtenv, but running which jupyter shows its in /usr/bin/jupyter, Ill try pip installing Jupyter in the virtenv and see if that resolves the issue.

[Homunculus Reticulli]

Nope, installing jupyter in my virtenv did not reolve the issue, I still get the error message: ImportError: No module named 'sklearn'

[Homunculus Reticulli]

Actually, I created a new virtenv (--no-site-packages) and pip installed the modules I wanted - now it works. Thanks

[ogrisel]

> ImportError: No module named 'sklearn' Then it actually did resolve the jupyter issue. You just need to pip install scikit-learn in your venv as well.

[Elliptica]

I can see the kernel find and connect to it in jupyter, but it can't find my modules still. Ideas?

[andilabs]

the answer of @ClimbsRocks Now run the kernel "self-install" script: python -m ipykernel install --user --name=my-virtualenv-name Replacing the --name parameter as appropriate. solved the problem for me

[cglacet]

Don't forget to switch kernel in jupyter.

[Valentas]

ipykernel install works, but unfortunately adds a new kernel globally, i.e. on sessions not run within this virtual env.

[cglacet]

Yep, that's a problem for sure.

[xtof54]

Installing jupyter in virtualenv is not working here: it fails to connect to the kernel

[Jed]

@climbsrocks yea, as Elliiptica has stated, I've done this and still not able import packages. Any ideas?

[Cornelius Roemer]

Amazing, this worked directly for me. In step 1 the --python flag was unnecessary for me (didn't know what to put there so left it out), also step 2 was redundant for me.

[MrR]

Where is 5 performed? In the activated VENV_NAME? why should I repeat step 4 every time I install a package??

[Roger Steinberg]

thats the real answer... the order is crucial to making this work.

[Kriticoder]

launching jupyter-notebook from the same terminal where my env was activated did the trick for me..thanks!!!

[chiceman]

If it's not working for you, try running python -m ipykernel install --name=my-virtualenv-name from within your activated virtual environment. Note: I removed the --user flag. This will install the kernelspec to the system directory. I'm not an expert on the topic, but I it seems to source the correct path when you launch jupyter-lab. Worked for me!

[jtlz2]

Step 4 makes this a golden answer - thank you so so much!! Highly bookmarkable.