Setting up a minimal data-science environment in Windows using WSL (part I)

code
linux
setup

Install Conda in WSL

This is the first part of a series of tidbits to set up a data-science environement in linux alongside windows, using windows subsystem for linux (WSL).

This will include conda as a package / environment manager, using the minimal version miniconda, and the following data-science packages installed for the default interpreter.

  • numpy
  • scipy
  • scikit-learn
  • matplotlib
  • seaborn
  • pandas
  • Jupyter

For deep learning packages, this comes next.

Install WSL

Install WSL: Open a powershell as administrator and run

wsl --install

Check the installation and version with

wsl -l -v

To activate the Bash shell type

wsl

To exit the shell type

exit

You can now use Bash to run linux tools and applications. During the installation process, make note of your login and password.

Accessing your folders and setting things up

Your home directory is accessed using cd ~. Create a Downloads directory and navigate to it:

cd ~
mkdir Downloads
cd Downloads

You can use explorer.exe . to open File Explorer from the current directory.

Update the Linux distribution

Run the following code to update the linux distribution you just installed and make sur it is up-to-date:

sudo apt update
sudo apt upgrade

Install python 3

Python comes pre-installed in WSL. Check the version:

python3 --version

If somehow it is not installed, you can install it by running

sudo apt install python3

Install miniconda

Even though WSL comes with python 3 pre-installed, we will use miniconda’s python distribution, which includes a package manager.

Go to https://docs.conda.io/en/latest/miniconda.html to download the installer of your choice (listed under Linux). At the time of this writing the latest installer includes Python 3.10, and put it inside your ~/Downloads folder, which is accessible from windows using the above-mentioned explorer.exe . command.

Place your downloaded file in this directory.

When you copy a file from Windows to a Linux directory in WSL, Windows may add a Zone.Identifier file to the directory containing the copied file. This is because Windows considers the Linux directory to be a network location, and adds the file to provide security information for the copied file.

The Zone.Identifier file is used by Windows to help protect against the execution of potentially harmful files that may have been downloaded from the internet or copied from an untrusted source. When a file is opened, Windows checks the Zone.Identifier file to determine if the file is safe to execute.

This may prevent you from installing the file you just downloaded. In this case, simply delete the Zone.Identifier file.

Make the script executable:

chmod +x your_conda_installer.sh

And run the bash script:

./your_conda_installer.sh

Follow the prompts, say yes to all. Quit WSL and start again so it recognizes the path to conda.

You’re done! Check your conda version using conda --version.