MDS software stack install instructions for Windows¶

Important

This guide has been (lightly) adapted from the UBC-Vancouver MDS Install stack under a CC-BY-SA 4.0 license.

Python, Conda, and JupyterLab¶

Python and Conda¶

We will be using Python for a large part of the program, and conda as our Python package manager. To install Python and the conda package manager, we will use the Miniconda platform (read more here), for which the Python 3.8 64-bit version can be downloaded here. After the download has finished, run the installer and accept the default configuration for all pages.

Do not add miniconda to PATH. We will set this up later.

After installation, open the Start Menu and search for the program called “Anaconda Prompt (miniconda3)”. When this opens you will see a prompt similar to (base) C:\Users\your_name. Type the following to check that your Python installation is working:

python --version

which should return something like this:

Python 3.8.3

If instead you see Python 2.7.X you installed the wrong version. Follow these instructions to delete this installation and try the installation again, selecting Python 3.8.

Integrating Python with the Git Bash terminal¶

To avoid having to open the separate Anaconda Prompt every time we want to use Python, we can make it available from the (Git Bash) terminal, which is what we will be using most of the time. To set this up, open the “Anaconda Prompt (miniconda3)” again and type:

conda init bash

You will see that this modified a few configuration files, which makes conda visible to the terminal. Close all open terminal windows and launch a new one, you should now see that the prompt string has changed to include the word (base) as in the screenshot below:

If you type

python --version

you should now see the same output as above:

Python 3.8.3

Note that if you want to run Python interactively from the Git Bash terminal, you need to prepend the winpty command, so the full command would be winpty python (if you run this, note that you can exit the Python prompt by typing exit()). Running just python works on other setups, but will freeze the Git Bash terminal.

Let’s also check the version of the conda package manager. If you type

conda --version

you should see something like this

conda 4.8.3

Optional One annoyance with our current terminal setup is that the word (base) is not on the same row as the rest of the prompt string (the part with your_name@your_computer. To fix this we can edit the .bash_profile configuration file to indicate that we do not want a newline at the beginning of the prompt string. Open up the configuration file using VS Code by typing the following command into a terminal:
code "/c/Program Files/Git/etc/profile.d/git-prompt.sh"
Delete the line that reads the following (it should be line 13):
PS1="$PS1"'\n'       # new line
Click to save the file, when VS Code prompts you that the saving failed, click “Retry as Admin” and then “Yes”. That’s it! Now if you launch a new terminal instance, you will see (base) on the same line as the rest of the prompt string as in the screenshot below.

Essential Python packages¶

conda installs Python packages from different online repositories which are called “channels”. A package needs to go through thorough testing before it is included in the default channel, which is good for stability, but also means that new versions will be delayed and fewer packages are available overall. There is a community-driven effort called the conda-forge (read more here), which provides more up to date packages To enable us to access the most up to date version of the Python packages we are going to use, we will add the more up to date channel, To add the conda-forge channel by typing the following in the terminal:

conda config --add channels conda-forge

To install packages individually, we can now use the following command: conda install <package-name>. Let’s install the key packages needed for the start of the MDS program:

conda install \
 "jupyterlab=2.*" \
 "numpy=1.*" \
 "pandas=1.*" \
 "flake8=3.*" \
 "black=19.*"
 "nbconvert=6.*"

conda will show you the packages that will be downloaded, and you can press enter to proceed with the installation. If you want to answer yes by default and skip this confirmation step, you can replace conda install with conda install -y.

Test Jupyter Lab setup¶

To test that your JupyterLab installation is functional, you can type jupyter lab into a terminal, which should open a new tab in your default browser with the JupyterLab interface. To exit out of JupyterLab you can click File -> Shutdown, or go to the terminal from which you launched JupyterLab and hold Ctrl while pressing c twice.

DATA 301