template-jupyter-project

Template to use for new python-based Jupyter notebook projects

First steps

This assumes that you are developing with conda and python 3.7, 3.8 or 3.9. These instructions should work for Windows when using Anaconda Prompt and for MacOS in Terminal (and by extension, likely will work on Linux).

1. Edit the environment.yml file:

• Uncomment the name attribute.
• After name, enter the name you want for your environment. Good practice is to use the same name for the environment as for the repo.
• Edit the dependencies list as required.

2. Create your environment:

conda env create -f environment.yml

3. If you want to make nice version-control friendly notebooks, which will remove all output and data upon committing, run

nbstripout --install

4. Examine .gitignore and decide where you want to put any large input or output datasets that you don't want to commit to GitHub, and fill in these paths.

Operation

Developing your package

Most of the time your workflow will look like this

conda activate your-env-name
jupyter notebook

Updating requirements

As you build the package you will likely want to add more dependencies. Edit the environment.yml file and run

conda env update -f environment.yml --prune

Please do not overwrite the environment.yml file using conda env export, as this exports everything in your local environment including all sub-dependencies and OS-specific packages (and sometimes local paths).