Public Policy Analytics

5

Geospatial Risk Modeling - Predictive Policing

DOI: 10.1201/9781003054658-5

5.1 New predictive policing tools

Of the few public-sector machine learning algorithms in use, few are as prevalent or as despised as the suite of algorithms commonly referred to as ‘predictive policing'. There are several flavors of predictive policing from forecasting where crime will happen, who will commit crime, judicial response to crime, as well as investigative outcomes.

Policing has driven public-sector machine learning because law enforcement has significant planning and resource allocation questions. In 2017, the Chicago Police Department recorded 268,387 incidents or 735 incidents per day, on average. Nearly 12,000 sworn officers patrol a 228 square ...