Forecasting Intimate Partner Violence to Inform Law Enforcement Interventions

• Activity Number: 214
• Type: Topic Contributed
• Date/Time: Monday, August 4, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): Richard Berk* and Charles Loeffler*+
• Companies: University of Pennsylvania and University of Pennsylvania
• Keywords: Intimate Partner Violence ; Forecasting ; Machine Learning

Abstract

The goal of the ongoing work is to provide, on a routine basis, forecasts of future incidents of intimate partner violence (IPV) that can help inform decisions made by Philadelphia police, prosecutors, and judges. The forecasts will be of perpetrator behavior. Using machine learning procedures shown to be effective in past work and data regularly collected by the Philadelphia Police Department, forecasting procedures will be developed from which usefully accurate projections can be made. The resulting software can be easily ported to local laptops and/or servers from which forecasts can be made as needed. The talk will provide a summary of progress on the project to date.