Cookiecutter Data Science

Intro

Why use this project structure?
Other people will thank you
You will thank you
Directory structure
Data is immutable
- data/raw and data/external are read-only folders!!!
- it doesn't need source control in the same way that code does
Notebooks are for exploration and communication
Keep secrets and configuration out of version control

Install cookiecutter
- pip install cookiecutter
- conda install -c conda-forge cookiecutter
- (linux only) apt-get install cookiecutter
- (mac only) brew install cookiecutter
demo
- set up the project
  - cookiecutter git@github.com:drivendata/cookiecutter-data-science.git
  - make help
  - make create_environment
  - make test_environment
  - conda activate my_project
  - make requirements
- git?
  - git init
  - create remote on GitHub
  - connect local to remote
    - git remote add origin git@github.com:pareyesv/my_project.git
  - git add .
  - git commit -m "chore: initial commit"
  - commit
    - git push -u origin master
  - package contents
    - import src help(src)
- Jupyter Notebooks
  - activate (pyenv env) (conda env)
  - install kernel
    - add ipykernel to requirements.txt
    - python -m ipykernel install --user --name my_project --display-name "Python (my_project)"
  - write
```
# file src/visualize/visualize.py
def hello():
  print("hello")
```
  - jupyter notebook example
```
%load_ext autoreload
%autoreload 2

import src.visualization.visualize as vis

vis.hello()
```
- documentation
  - cd docs/
  - make html
  - try the RTD theme
  - autodoc
    - add .. autoclass:: src.visualization.visualize.Greetings
    - add extensions = ['sphinx.ext.autodoc', 'sphinx.ext.napoleon']