AUT's Centre for Social Data Analytics (CSDA) has presented findings of its Allegheny Homelessness Tool at the Bloomberg Data for Good Exchange (D4GX) in New York.
The Bloomberg forum aims to build cross-sector relationships while solving problems for the social good that might not otherwise be addressed by market forces.
CSDA Director, Professor Rhema Vaithianathan, and research fellow Dr Chamari Kithulgoda attended in September and presented a detailed poster – “Implementing a Predictive Risk Model to Prioritize Homeless Services: The Allegheny Homelessness Tool”.
The poster, created with Allegheny County Department of Human Services, traces how CSDA has assisted the county to leverage its integrated data in an effort to replace its existing actuarial tool for homelessness services with a predictive risk model (PRM) that better identifies clients at risk of harmful outcomes.
The authors conclude that “Overall, the PRMs performed well, even on harms they were not trained for. By contrast, the scores of the actuarial tool were only weakly related to the future harms that we studied.”
Implementation of the PRM tool will provide better targeting of restricted homelessness services to those who are most likely to be subject to future harms. It will also reduce the amount of time that DHS staff spend assessing clients.
The PRM tool is expected to be deployed by early 2020.
Data for Good Exchange (D4GX) is an annual event that brings together data scientists with non-profits and government agencies to solve some of the world’s biggest problems.
Participants share insights and progress on applying modern machine learning and data science methods to problems in public and non-profit sectors as well as methods around their application, data collaboration and data science education.