Building a Deeper Police Early Intervention System

Fellows: Thomas Ralph Davidson, John Henry Hinnefeld, Edward Yoxall
Data Science Mentor(s): Jen Helsby, Joe Walsh
Project Manager: Allison Weil
Project Partner: Charlotte-Mecklenburg Police Department

In 2015, DSSG participated in the White House Police Data Initiative, partnering with the Charlotte-Mecklenburg Police Department (CMPD) to build a better “early intervention system” (EIS). These systems help police departments determine which officers are at risk of an adverse incident — such as a sustained citizen complaint, excessive use of force, or on-duty injury — so that they can receive additional training and assistance before these events occur. Our project in 2015 used anonymized data about officers, arrests, complaints, and other local measures to produce a new model for CMPD that predicts risk with more true positives and fewer false positives than their current system.

In 2016, we continued our partnership with CMPD to further refine the EIS, improve results and added additional detail to the system. For example, we developed the model’s dispatch-level predictions and identified factors on a given shift that may raise the potential for a subsequent adverse incident. The team also worked with the Nashville DSSG team on building a more robust EIS that can be modified to work with data from additional police departments in the future.

For updates about this project, our work with the Metropolitan Nashville Police Department, and our future plans for improving and replicating EIS, please see our blog.