ML Project Template

cookiecutter template for DS projects

Inspired by this repo

Requirements to use the cookiecutter template:

• Python >= 3.5
• Cookiecutter Python package >= 1.4.0: This can be installed with pip by or conda depending on how you manage your Python packages:

$ pip install cookiecutter

or

$ conda config --add channels conda-forge
$ conda install cookiecutter

• dvc >= 0.93 (optional, only if you want to control your data using DVC)

To start a new project, run:

cookiecutter https://github.com/analysiscenter/ml_project_template

You will be asked for these fields:

Field	Default	Description
project_name	Template	Verbose project name, used in headings (docs, readme, etc).
repo_name	project_name lowercased with spaces replaced by underscores	Repository name on GitHub (and project's root directory name).
include_library	no	Chose a library to attach as a git submodule. no means that no library will be attached.
git_config_user_name	do_not_set	Specify local user name for git. Default is to skip this.
git_config_user_email	do_not_set	Specify local user email for git. Default is to skip this.
init_DVC	yes	Select whether you want to initialize DVC in the project.
dvc_cache_dir_mode	pass if you selected not to initialize DVC, else infer	Select cache directory for DVC. If pass, nothing is done; if infer, path is selected as /data/<repo_name>/.dvc/cache depending on include_library, if custom, value from custom_dvc_cache_dir is used
custom_dvc_cache_dir	pass	Provide a path to DVC cache directory, if you selected custom as dvc_cache_dir_mode.
lib_dvc_cache_dir_map	default	Just press Enter :) This sets some utility values.

GitHub repositories should be created manually, but automatic GitHub repository creation can be added in future

The resulting directory structure

The directory structure of your new project looks like this:

.
├── datasets                  <- Keep your datasets here
├── docker_containers
├── .dockerignore
├── .dvc
├── .dvcignore
├── extra                     <- extra helper utilities that are not project-specific, ex. cookiecutter template updater
│   ├── .cookiecutter.json
│   ├── src
│   └── update_cookiecutter_template.sh   <- run this script to merge recent changes in cookiecutter template into your project
├── .git
├── .gitattributes
├── .github
│   └── workflows
│       └── status.yml
├── .gitignore
├── .gitmodules
├── notebooks                 <- Development notebooks
├── overview                  <- Notebooks with overview of main results
├── pylintrc
├── readme.md                 <- The top-level README for developers using this project.
├── requirements.txt
├── src                       <- Project-specific models and utilities
│   ├── Library submodule     <- SeismicPro, SeismiQB, or batchflow as a git submodule
│   ├── __init__.py
│   └── readme.md
└── tests