Data-Science-template

A starter template for data science work using Pixi. Created by Brian Mc Donald

Replace this file with a README specific to your project

Getting started

1. Install Quarto, the tool/format for publishing to different formats
2. Install Pixi to manage packages. On MacOS curl -fsSL https://pixi.sh/install.sh | bash and on Windows Powershell: iwr -useb https://pixi.sh/install.ps1 | iex
3. Run pixi init . to create a new Pixi project. It creates a pixi.toml and pixi.lock file.
4. Run pixi add <packagename> to add pacakges. Example: pixi add python ruff polars matploltib requests
5. Do exploration in notebooks - ipynb's will keep cell output, which is good for exploratory analysis but is messy for version control.
6. When code is maturing, move it to files in src. The examples can be changed as needed.
7. Create quarto files for the analysis and formats you want, calling the code from src - dashboards, websites, pptx pdf.
8. Run quarto render in the quarto directory to render the quarto files and check that the outputs appear in the Reports directory.
9. If you wish to publish a directory as a static site, you can use this Github Action to copy a directory to a gh-pages branch, which you can then configure using Github Pages.

Tips

• Run ruff format to format all files in a directory to keep your code clean. The Ruff VSCode extension is convenient.
• Use .gitignore to exclude certain files and directories from version control. Definitely exclude .DS_Store, .pixi, .env. Consider excluding data, reports and notebooks.
• Use Pixi tasks to simplify running commands or scripts. Example pixi task add cow cowpy "Hello User" or pixi task add fetch python data_get.py --option all, then running pixi run cow or pixi run fetch
• Use pixi update or pixi upgrade to update package versions

Structure