Created this template to make machine learning projects more structured. As a reference I based my template on the cookiecutter-data-science structure.
- Install python-poetry (https://python-poetry.org/docs/#installation)
- Execute the setup.sh-file
bash setup.sh
|-- data/
|-- raw/ <- The original, immutable data dump.
|-- processed/ <- The final, canonical data sets for modeling.
|-- documentation/ <- Textual documentation
|-- notebooks/ <- Trained and serialized models, model predictions, or model summaries
|-- models/ <- Jupyter notebooks. Naming convention is a number (for ordering),
│ the creator's initials, and a short `-` delimited description, e.g.
│ `1.0-jqp-initial-data-exploration`.
|-- src/ <- Source code for use in this project.
|-- config.py <- Set environment variables.
|-- add-jupyter-venv.sh <- Adds virtualenv to ipykernel.
|-- pyproject.toml <- Installs project and sets up environment.
|-- .gitignore <- gitignore.