Predicting Substance Abuse Treatment Outcomes

Team Recovery (chiwang-nkdunbar-rckywong-stevenzw)
Francis Ng <[email protected]>
Nicholas Dunbar <[email protected]>
Ricky Wong <[email protected]>
Steven Wang <[email protected]>

University of Michigan School of Information (UMSI)
Master of Applied Data Science (MADS)
SIADS 697 - Capstone
Winter 2022

Key links

• Project home
• Full report
• GitHub repository

About this repository

This repository contains work related to render this README and our project report, as well as all relevant notebooks, datasets, and utilities.

Repository structure

dataset/

Copies of the Treatment Episode Data Set: Discharges data (2016, 2017, 2018, 2019)

notebooks/

Jupyter Notebooks (.ipynb files) used for data ingestion, cleaning, training, modeling, prediction, and visualizations.

All notebooks are intended to be able to run start-to-finish, independently, in any order. Dependency installation is done via a library import.

More details within notebooks/README.md.

notebooks/lib/

Convenience functions used among multiple notebooks to do multiple things. Includes the installation of requirements.txt packages.

report/

Our detailed findings in a write-up, intended to be generated and served off of GitHub Pages. More details within report/README.md.

Tested environments

• Coursera's Milestone 2 (SIADS 694-695) (Jupyter) notebook environment
  • 4 CPUs, 16GB RAM, 0 GPUs
  • Python 3.7.10
  • jupyter-core 4.7.1
  • jupyter-notebook 6.2.0
• University of Michigan's Great Lakes cluster Jupyter Notebook (python3.8-anaconda/2021.05)
  • 4 CPUs, 16/32/64GB RAM, 1 GPU
  • Python 3.8.8
  • jupyter-core 4.7.1
  • jupyter-notebook 6.4.3