About the model

The predictive policing software currently being tested in Santa Cruz, CA (and in Los Angeles starting Fall 2011) is based upon a self-exciting point process model of crime.  The model estimates the risk of “aftercrimes,” crimes triggered by recent events due to repeat offender behavior.

About the software

Each day Santa Cruz Police Department uploads the latitude, longitude, estimated time, and crime type for all residential burglary, auto burglary, and auto theft events occurring between 2006 and the current day (approx. 4500 events).  The highest risk areas (150m x 150m cells) for that day are then computed and displayed on a google map.  Patrol officers are provided with maps and directed patrols are made throughout the day to these hotspot areas on officer free patrol time.  A key feature of the experiment is that, because crime is dynamic, hotspot areas can change on a daily basis.  Police patrols therefore must adapt to this dynamic risk.

For updates on the experimental predictive policing software

We currently are testing the software and exploring how it can be made more widely available.  If you are a member of a police agency or other party interested in eventually obtaining the software, please fill out the form located here.