I’m a Professor in Electrical and Software Engineering in the Schulich School of Engineering at the University of Calgary, Canada. This site describes my work with community, university research and teaching.

University Contact Information:
2500 University Drive N.W.
Calgary, Alberta, Canada
T2N 1N4
Phone: +1-403-210-8976
Email: gmessier@ucalgary.ca
Office: Room 308, Information and Communication Technology (ICT) Building

Housing, Homelessness and Data

My research and work with community involves using data science to improve the services provided to people experiencing homelessness, mental health disorders and addiction. This involves work with cross-disciplinary teams that include community agency staff and researchers from medicine and social work. I’m proud to collaborate with some amazing agencies, including The Calgary Drop-In Centre, the Canadian Alliance to End Homelessness, the Calgary Homeless Foundation and the Trellis Society.

Here are a few presentations that provide an overview of some of my work.

• How to Recruit University Data Experts to Help Your Agency
• Can Chat-GPT Style AI Help Us Understand Emergency Shelter Data?
• Spreadsheets vs AI: When Good Enough is Good Enough
• How to Link Housing/Homelessness Agency Client Records and Preserve Privacy
• Visualizing Journeys Through the Homelessness and Housing System of Care

Looking for Data?

I do a lot of work with the administrative data recorded by agencies each time a person accesses housing or homelessness services. If you’re curious about doing similar work, it’s nice to quickly get started with a data set. While some anonymized medical administrative data sets exist in the public domain (eg. MIMIC), there’s nothing similar for housing and homelessness that I’m aware of.

To help fill this gap, I’ve created a synthetic administrative data set that contains fictional records of how people interact with a housing and homelessness system of care. This data does not contain real people so there are no privacy concerns and it is statistically similar to real Calgary housing and homelessness data.

The data set can be downloaded here. The page includes a link to the software I wrote to generate the data set. This code can be reconfigured to generate data sets that represent different sized communities. There’s also a link to the technical paper explaining how the synthetic data model works and evaluating how well its statistics match real data.

WARNING: The synthetic data statistics do not accurately represent the exact experience of real people in specific communities. For example, if you want to know the actual homelessness re-entry rate in a particular community, you must use real data from that community. The synthetic data should also not be used to train machine learning/AI models that will be used to deliver services to real, non-synthetic people. This data set is meant as a sandbox for prototyping and brainstorming only.

AI Guiding Principles

As a data scientist, I do sometimes apply AI and machine learning to the problems I work on. AI is evolving quickly and I think all researchers who use it should have some guiding principles to anchor their work. Here are mine:

1. I will not work on a project that involves developing an AI interface (chat or otherwise) to interact directly with a person in crisis.
2. I will not work on an AI project meant to replace a human decision maker or care giver.
3. I will use AI to improve the quality, intepretability and availability of the information that human care givers require to make the best possible decisions for the people they help.

A Note about Language

Language matters and it’s important to use terms that do not further disadvantage the people we’re trying to help. In particular, the use of the term “vulnerable people” can suggest that vulnerability is a property of that group rather than a set of external circumstances they are experiencing, many beyond their control. An excellent discussion of this issue that can be found here.

I am committed to not perpetuating harm with the language I choose. However, this has been a journey for me and some of the language I use in older publications and presentations may fall short of current best practices.

Teaching

On my YouTube Channel, you can view my lectures on:

• Computer Architecture (2nd year undergraduate course)
• Computer Networks (4th year undergraduate course)
• Electronics (2nd year undergraduate course)
• Probability and Stochastic Processes (graduate course)
• Wireless Communications (graduate course)
• Digital Communications (graduate course)

You can also view a list of the classes I’ve taught in the past here.