cookiecutter-ds-docker

A Docker-based Data Science cookiecutter (for myself)

Table of Contents

• 1. Introduction
• 2. Documentation
• 3. Quickstart
• 4. Known Issues
• 5. License
• 6. Authors

1. Introduction

This repo hosts a personalized, Docker-based cookiecutter template for Data Science projects.

The template consists of a docker-compose stack with the services below:

• A customized Jupyter service with a starter Python package installed.
• An mlflow tracking server to store experiments
• A postgresql database, which stores mlflow tracking information

• Python test and development
• Building Sphinx documentation

2. Documentation

Please refer to Read The Docs for the documentation.

3. Quickstart

First, install cookiecutter and docker.

Then, run the commands below and follow the on-screen instructions to "cut" a new project from the template:

cd /{ base_folder }
cookiecutter gh:sertansenturk/cookiecutter-ds-docker

To build and start the Docker stack, run:

cd /{ basefolder }/{{ cookiecutter.reposlug }}
make

4. Known Issues

• In Linux, if DATADIR (defined in the template) is in a drive formatted in NTFS, you might have permission issues when mounting the folder to the docker containers (See Issue #56). We suggest you to cut the project into a drive, which is formatted in a native Linux filesystem.