Last week, the Vancouver Police Department (VPD) released their findings from a 6-month program they undertook with us in 2016 to pilot a new project, a state-of-the-art machine learning system that helps predict when and where crimes may occur.
The goal of VPD’s pilot program was to prevent break-ins and property crime throughout metro Vancouver, which typically spikes during the summer months when many people are away on vacation or leaving doors and windows open to beat the heat.
VPD focused on areas south of Broadway where residential break-ins are the most common. Special Constable Ryan Prox reported that their program reduced property crime by as much as 27% in some areas (compared to the past 4 years).
The technology turns simple, historic point crime data into powerful, actionable crime predictions. Predictions are quite precise; the system forecasts within a 100 m X 100 m (300’ X 300’) block, with a 2-hour time window, to determine where and when a particular type of crime might occur. By reviewing what took place in the field against predictions made for the same day, Vancouver’s Chief of Police said the system predicted with up to 80% accuracy.
With their successful pilot program complete, VPD has rolled out the solution to the entire police department: the first in Canada to adopt a “predictive policing” system.
To learn more about VPD’s program, check out the articles below:
If you want to learn more about how Geocortex can be used to help keep communities safer, we’d be happy to chat! You can contact us here.
Contributors