Switchbox Copier Template

Documentation: https://switchbox-data.github.io/data-science-project/

This Coper template lets you bootstrap python packages or polyglot data science projects.

For python packages, it supports:

• Modern python tooling: uv, ruff, ty, pytest, tox
• Package building and publishing with hatchling
• MkDocs documentation with Material theme

For python data science projects, it supports:

• Pydata libraries: polars, pyarrow, seaborn
• Quarto notebooks
• Modern python tooling: uv, ruff, ty

For R data science projects, it supports:

• Tidyverse libaries: ggplot, dplyr, etc.
• Quarto notebooks
• Modern R tooling: pak, air, radian
• Fast install of compiled R packages via pak and P3M
• VS Code extensions for R

All project types include:

• just for task automation and project management
• devcontainers for consistent development environments
• pre-commit hooks for automated code quality checks
• Optional AWS integration (boto3 for Python, AWS CLI for all)
• Optional GitHub Actions workflows for CI/CD

Using the Template

To create a new project from this template:

# Install copier
uv tool install copier # or use pipx

# Create a new project (requires --trust for git initialization)
copier copy --trust https://github.com/switchbox-data/data-science-project.git path/to/my-new-project

You'll be asked a series of questions to customize your project, including whether you use Github or AWS, and the type of project you want to make:

• Python Data Science: Includes polars, pyarrow, seaborn, numpy, Quarto notebooks
• R Data Science: Includes R, renv, Quarto, R extensions
• Python Package: Includes build tools, twine, MkDocs documentation

You can select one or more of these project types! They are discrete modules that compose neatly. And you can come back and add a module later.

After making your selections, copier will:

• 🔧 Initialize a git repository
• 📁 Add all generated files to git
• 💾 Create an initial commit
• 🚀 Set up your development environment

Using the CLI

If you want to skip copier's interactive prompts and generate the project directly, you can pass arguments directly.

For instance, the following prompt generates a python package + data science project:

copier copy --defaults --force \
  --data project_features='["python_data_science","python_package"]' \
  https://github.com/switchbox-data/data-science-project.git /path/to/new-project

See copier/copier.yml for all questions and defaults.

Development

• Install dev deps with uv:

uv sync --group dev

• Run tests (pytest runs copier to generate scenarios and asserts outputs):

uv run pytest -q

• Lint/format:

uv run pre-commit run -a

• Serving the docs locally

uv run mkdocs serve

License

MIT