Skip to content
Christopher Aicher edited this page Mar 8, 2017 · 1 revision

Overview

The suggested project layout is as follows:

  • data/: folder containing raw, processed, and output data
  • docs/: folder containing written notes and papers
  • experiments/: folder containing experiment scripts
  • notebooks/: folder containing notebooks for visualizing results
  • src/: folder containing source code

Data Folder

The data/ folder contains raw data, processed data, and experimental output.

The organization is by date (data/MM-DD-YYYY/) and then by type (data/MM-DD-YYYY/<type>). <type> is either input for processed input data or output for experimental output. Raw or third-party data is in a separate folder labeled data/raw/

Documents Folder

The docs/ folder contains working paper drafts and notes.

Experiments Folder

The experiments/ folder lists all the experiments scripts used.

The organization is by date-name (experiments/MM-DD-YYYY-brief-name).

Each experiment folder has a master_script.sh file that will execute the experiment when called from the project root directory.

Notebooks Folder

The notebooks/ folder lists all output data analysis and figures.

The organization is by date-name (experiments/MM-DD-YYYY-brief-name).

Source Code Folder

The src/ folder contains all the source code used in experiments.

See src/README.md for installation instructions. This typically involves creating a conda environment and installing source modules/packages.

Version Control

The following files are ignored from version control:

  • The entire data/ folder
  • Any temporary files in a temp folder are ignored (usually for experiments)
  • Temporary LaTeX and Knitr output in docs/ and notebooks/