Each assignment is organized and its Dockerfile and Python code is provided in its own directory in the repository. Assignment4 is used as an example; other assignments follow the same pattern.
Dockerfile, which the user can run to create an image and run code.Assignment4.py, which is the main Python script and which may import other local Python modules.Assignment4.ipynb, which is a Jupyter notebook that the user can use to interactively run code in the.pyfile.readme.md, which provides an overview of the assignment's purpose and instructions for running associated code.
Input data files for each assignment are provided in the data directory in this repository. Each assignment has its own directory in data and each assignment has an input directory, raw, and an output directory, processed. Some assignments may include intermediate outputs, which are saved in interim. Inputs are immutable - they're never changed. Code in this repository is idempotent, meaning if the user runs the same code again outputs will always be the same (notwithstanding differences owing to intended stochasticity / randomness in the code).
Rather than clone the repository to run the code, the user can opt to pull Docker images, tagged by assignment, from the associated DockerHub repo. Pull the appropriately tagged image to run the desired assignment program. For instance, to run code for Assignment 4, execute $ docker pull pgrjhu/705.603:module4-1.1.
Additional instructions for running an image are provided in the DockerHub repo README.