Quantifying Relationships: Leveraging Data Science for the National Network’s Intimate Partner Violence Intervention (IPVI)

• In Event: Machine Learning and Causal Inference

Abstract

This paper discusses the current, flawed, state of intimate partner violence (IPV) crime data and an approach to addressing this issue in the context of a focused deterrence strategy to control dangerous offenders and prevent victimization. Intimate partner violence is a serious public health issue from both a criminal justice and societal perspective. It is pervasive and deeply personal, and the full scope of IPV occurring in most jurisdictions is unknown for predominantly two reasons: IPV often goes unreported, and what is reported is often documented in ways that make it difficult to evaluate its prevalence as a distinctive type of crime. Historically, while limited data on IPV existed, it varied from jurisdiction to jurisdiction and seldom existed in a format that allowed for easy analysis. Utilizing data science and machine learning tactics, the NNSC has been able to develop a method that allows for jurisdictions to extricate IPV from the broad category of DV offending in an accessible manner. This enables jurisdictions to fully understand the nuance and depth of their local IPV offending dynamics, and develop strategies to effectively address the violence in their community on a personalized level.

Authors

• Kyle Ott, National Network for Safe Communities / John Jay College

• Rachel Teicher, National Network for Safe Communities