This project lays the groundwork for developing predictive models and improving work processes in social agencies that serve underserved populations. The specific targets of the work are long term homelessness and recidivism (tendency to relapse), with a primary focus on the city of Boston, Massachusetts in collaboration with a community partner (Pine Street Inn). The planned integrative research includes: predictive modeling to assess needs of the homeless guests to improve the match against the right services, and process modeling to improve triage work at the homeless shelter. The project will identify data necessary for the modeling efforts and resolve data-related problems such as data quality and completeness; and assess current triage practice to develop and experiment with new versions of work processes at a homeless shelter.

This project will advance current knowledge about work practices and service matching at homeless shelters, and improve understanding of data availability and quality to support these outcomes. The work will develop ways to better serve the homeless population as well as lay the foundations to roll out the models and results to other communities in Massachusetts as well as other states and other underserved populations, such as at-risk families, populations suffering from disabilities, and those experiencing substance abuse or addiction. Significant societal benefits can accrue through improved and efficient support for the homeless, either directly through services or indirectly through better use of federal and local taxes. Providing services that match the needs of the homeless, and improving the work practices at the shelter can reduce both the length of stay and recidivism, contributing to the overall quality of life within society.