A cookiecutter template for a data science project file structure following Microsoft's Team Data Science Process (TDSP) guidelines. This template allows for the easy creation of a full project directory that is both language- and cloud-agnostic.
- Install the project template utility cookiecutter using the package manager of your choice.
$ pip install cookiecutter$ conda install cookiecutter$ sudo apt-get install cookiecutter
- Generate your project using the project template from this repository. You will be prompted for both the project name and the customer or client.
$ cookiecutter https://github.com/pmason314/cookiecutter-tdsp
project_name [Placeholder-Data-ScienceProject]: my_data_science_project
customer_name [Placeholder Customer Name]: John Smith
- (Optional) Initialize a git repository in the newly created project directory. You can choose to take advantage of the provided .gitignore file, which contains the GitHub default for Python- and R-based projects.
$ cd my_data_science_project
$ git init
Each subdirectory contains a README (omitted from this diagram) containing a template and recommended contents for that stage of the data science life cycle.
[Project Name]
├── code
| ├── exploratory_data_analysis
| ├── modeling
| └── deployment
├── sample_data
| ├── raw
| ├── processed
| └── for_modeling
├── Docs
| ├── Data Dictionaries
| ├── Data Reports
| ├── Model
| | ├── Baseline
| | └── Model 1
| └── Project
| ├── Charter.md
| ├── Exit Report.md
| └── System Architecture.docx
├── README.md
└── .gitignore
Microsoft's Team Data Science Process is an agile, iterative, data science methodology to improve collaboration and team learning. It is supported through a lifecycle definition, standard project structure, artifact templates, and tools for productive data science. The original archived repository containing the TDSP format is available at https://github.com/Azure/Azure-TDSP-Utilities.