Documentation Status Project License GitHub issues Version RTD Version Astropy

Structuring your Project

Author: Victor Calderon (

Description: A logical, reasonably standardized, but flexible project structure for doing and sharing data science work.

Now that your have a working version of python on your computer, you can start doing research.

One of the key elements of a project is for it to be reproducible by others. Having this in mind when you’re structuring your project will allow others to look at your code, understand it well enough to be able to recreate your results.

This is a short guide on 2 ways to structure your code, without having to do much of creating documets, etc.

Cookiecutter and Folder structure

Cookiecutter is a command-line utility that creates projects from cookiecutters (project templates), e.g. Python package projects, LaTeX documents, etc.

Cookiecutter has been widely used for many projects, and each team and organization can create their own template. For more information, visit the cookiecutter documentation.

As the famous say goes:

Don’t reinvent the wheel!

You can always create your own folder and file structures, and organize your documents the old-fashioned way. The problem with this is that it may vary from project to project, and it will be more difficult to be consistent and effective throught your projects.

For this reason, I rely on cookiecutter templates to create the file and folder structure of a project.

There are many different cookiecutter templates out there, but after trying to find the best one that suits my needs in research and programming, I found one that works great! And after some modifications, I came up with a version of this template.

These two templates are shown in Data Science - Cookiecutter and Personal version - Cookiecutter. But first, let’s make sure you have cookiecutter installed correctly.

Requirements to use cookiecutter templates

The minimum rquirements for creating cookiecutter templates are:

  • Python 2.7 or 3.5
  • Cookiecutter Python package >= 1.4.0: This can be installed with pip or conda depending on how you manage your Python packages.

You can install it by typing this on the terminal

pip install cookiecutter

or via Anaconda:

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

Now you can use cookiecutter to create new templates for projects and papers!

Data Science - Cookiecutter

Cookiecutter Data Science is best described as

A logical, reasonably standardized, but flexible project structure for doing and sharing data science work.

This folder structure allows everyone looking at your code to understand it right away. It also provides many different functions (as part of a Makefile) that simplify the workflow of your project.

In a nutshell, this cookiecutter includes:

  • A Makefile file with **useful functions.
  • Documentation to make your project easily accessible and readable
  • And more!

In order to use this template, you follow the documentation in Cookiecutter Data Science.

Personal version - Cookiecutter

If you need more than the normal Data Science Cookiecutter template, you can use my version. Some of the differences are:

  • It includes and easy-to-use environment.yml file that makes it easy to install dependencies.
  • Extra functions in the Makefile.
  • Choice of what kind of documenation to use. One has the option choose from traditional Read The Docs style or the Astropy Sphinx Theme.

You can check how these two styles look like:

  • Version RTD - Read The Docs Version
  • Version Astropy - Astropy Version

Next, you can create your own Project based on this cookiecutter version

To start a new project and prompts

To start a new project, type the following:

$ cookiecutter

If you want the default project scheme from DrivenData (see above), run:


Depending on what kind of folder structure you want, you might want to choose from the different types.

After running this command, you will be prompted some questions regarding the parameters for the project. This will prompt you to answer a few questions like:

Question Description

Name of the project. This can be similar to one on Github.


  • SDSS_analysis
  • Lung_cancer_analysis

Name of the directory/repository, the project will be saved. This field should not contain spaces should not contain spaces


  • Calderon_Victor_Astro_PhD_Thesis
  • Szewciw_Adam_Astro_PhD_Thesis

Author’s first name. It can include spaces


  • Adam Sanchez
  • Rose Roserberg

Author’s email address.



A short description of the project. This can be a brief overview of the project.


  • Repository for lung cancer analysis
  • Analysis on galaxies and cosmology

A longer version of short_description


  • Repository for lung cancer analysis
  • Analysis on galaxies and cosmology

Type of License for the paper. Without this, one cannot use any of.


  • MIT
  • BSD 3-Clause
  • GNU GPL v3+
  • Apache Software Licence 2.0
  • BSD 2-Clause*
s3_bucket Path to AWS storage. This is temporarily disabled!!
aws_profile AWS profile name. This is temporarily disabled!!

Name of the project’s anaconda environment. In order to use the packages of this project, you need to first activate this environment.


  • lung_cancer_env
  • SDSS_galaxies_env

Author’s Github username. This will be use tolink the project to the Github repository.


  • username
  • username2018

Name of the project on Github


  • Awesome_lung_analysis_username_2018
  • Another_awesome_analysis

Version of the project.


  • 0.0.1
  • 0.1.2dev

If ‘y’, ‘.travis.yml’ will be installed This is useful when doing ‘continuous integration’


  • y
  • n
use_read_the_docs If ‘y’, it will use ReadTheDocs for docs

URL of the project on Github! You need to create this repository separately.*


minimum_python_version Python version to use for this analysis. This will set the python version in the ‘environment.yml’ file.

It assumes Python 3 version by default


  • Adam
  • Rose

Option for which kind of documentation to use. See the examples above to see which one you like.


  • RTD: Used ReadTheDocs-type of docs
  • Astropy: Used Astropy-type of docs

Using the Template

Now that one has answered the questions from To start a new project and prompts, you just need to populate the project with scripts, notebooks, and of course, documentation!!

The structure of the finalized project can be found in the Project Structure section.

Editing your environment

Now that you have a working proect from cookiecutter, you can start by editing the environment of your project.

If you downloaded my version of cookiecutter, you should be able to edit the environment.yml file. This file states which packages need to be installed by Anaconda and pip in order to run the scripts of the package.

The environment.yml file looks like the following:

name: name_of_environment

  - defaults

  - python>=3.6
  - ipython
  - anaconda
  - astropy
  - h5py
  - numpy
  - pandas
  - scipy
  - seaborn
  - pip
  - pip:
    - GitPython
    - progressbar2

You can edit the environment.yml file to include/exclude packages needed by your project.

After having edited the list of packages needed by your project, you can execute the command

$ make environment

to create the environment.

If you have done this step before, and you want to update the environment, you need to run

$ make update_environment


Adding your Project repository to Github

If you follow the instructions from above, you should have

  • Downloaded the repository
  • Created your own project with the desired file and folder structure
  • Created your working environment for you project

The next step is to add it to Github and make it accessible.

To do this, your should do the following:

  1. Create a Github repository with the same name as the repository.
  2. Type git add remote origin<username>/<project_name>.git. In here you need to replace <username> and project_name with your details.
  3. git push origin master - This will push your project to Github.

To check that you did this correctly, type

git remote -v

and you should get something that looks like this:

origin<username>/<project_name>.git (fetch)
origin<username>/<project_name>.git (push)

where username and project_name pertain to your repository on Github.

Now all of the files are online on Github, and should be ready to integrate them with Read The Docs.

Documentation for your new project

Now that you have both a working local and online copy of your code, the next step is to create the documentation for the project.

For this, you can easily use Read The Docs (RTD).

You need to do the following:

  • Create an account on “Read the Docs”
  • Go to your Profile and select My Projects
  • From there, you should import the repository manually (it’s easier). Click on Import a Project and follow the instructions.
  • You should add the project with the same name as the Github Repo if possible. Otherwise, you might need to change the links to the badges on the files in the project, among others.
  • Make sure that the repository was correctly built by looking at the Builds and see that it compiled correctly. If not, it should tell you if there was an error and what the error was.
  • Now you go and change the documentation depending on the project’s needs.

Continuous Integration for your Project

Continuous integration deals with testing your code for possible errors, and making sure that everything is working as expected. Depending on your project’s needs.

This template includes a .travis.yml, which the files used by Travis CI. Travis CI is a Continuous integration platform for testing your code, and checking the functionality of your project.

Project based on the modified version of the cookiecutter data science project template.