Software Engineering Practices for Data Science¶
Before (or alongside) either level, it's worth picking up the habits that keep research code reliable: dependency isolation, keeping credentials out of version control, and a standard project layout. The UF AIBHS program maintains a full walkthrough of this. This page just points at it rather than duplicating it.
Source: uf-aibhs.github.io/extra-resources/software-engineering
- Virtual Environments: Isolating Project Dependencies
- Keeping Secrets Out of Git: .env & python-dotenv
- Git & GitHub Desktop: Version Control from Scratch
- Cookiecutter Data Science: A Standard Project Skeleton
- Science as Amateur Software Development (talk)
Read in that order, before starting a Foundations, Clinical Application, or Basic Science project:
- Dependencies
- Secrets
- Version control
- Project structure
For the deeper version of the same argument, see Better Code, Better Science.