Reducing Recidivism with Community-based Programs: Data-driven Analytics for Societal Benefits and Fairness
Pengyi Shi
Associate Professor
Mitchell E. Daniels, Jr. School of Business
Purdue University
Abstract: Recidivism -- a formerly incarcerated individual commits a new offense -- is one of the most challenging issues facing the modern criminal justice system. Notably, many recidivated individuals suffer from mental health and substance use issues, yet they are often criminalized and incarcerated for minor offenses instead of receiving community-based assistance and treatment. Incarceration-diversion programs have been advocated to break this vicious cycle but lack analytics-based decision support. We work with community corrections and diversion programs to study the complex process flow of individuals through the criminal justice and social services support systems. We build MDP and queueing models to capture key tradeoffs among jail overcrowding, reoffending, and in-program violations. We develop data-driven analytics based on machine learning and reinforcement learning to aid the allocation of limited treatment resources to maximize the benefit to society while reducing racial disparity in incarceration and recidivism.
Bio: Pengyi Shi is an associate professor at the Mitchell E. Daniels, Jr. School of Business, Purdue University. She received her Ph.D. degree in Industrial Engineering from Georgia Institute of Technology before joining Purdue in 2014. Her research interests include data-driven modeling and decision-making in healthcare and service operations. She has collaborated with practitioners from different healthcare organizations, including major hospitals in the US, Singapore, and China. Most recently, she is collaborating with community correction programs to develop data-based evaluation and human-in-the-loop machine learning algorithms. Her research has won the first place of IISE Outstanding Innovation in Service Systems Engineering Award in 2023, the first place of MSOM Responsible Research in OM Award in 2021, the first place of INFORMS Pierskalla Best Paper Award in 2018, and a finalist of the INFORMS Daniel H. Wagner Prize in 2023.
