HMIS Data Science Study

« back to Projects

HMIS Data Science Study

HMIS Data Science Study

Members of the Data Science Working Group at Code for San Francisco have been charged with answering the Community Technology Alliance’s prompt about homelessness programs.


What variables best predict whether an individual is categorized as ‘in permanent housing’ as an outcome, by population segment:

Veterans Chronically Homeless Continuously Homeless Has Disabling Condition Domestic Violence Victim Male/Female Latino/Non-Latino Data

Data is in HMIS format, a data standard defined by the US Department of Housing and Urban Development


View the HMIS Data Science Study Presentation for a summary of our findings

Featured Notebooks

Load and clean the data Explore the data Feature engineering to prepare input variables Make outcome predictions with logistic regression model Setup

Install Jupyter Notebook; this is most easily done by installing Anaconda:

Install seaborn. To do this in a new conda environment: conda create –name datasci seaborn

To deactivate/activate the environment: source deactivate datasci source activate datasci

Get Started

Fork this repository and clone it locally. Locate the dataset (pinned in #datasci-homeless on Slack). Run jupyter notebook Navigate to notebooks/load_data_example_v2.ipynb to start exploring the data. Additional information on completed and open items can be found in the pinned documents in #datasci-homeless.

Project Status beta
Skills Needed none right now
Tags inferential statistics regression analysis homelessness data science
Edit Project