Utility of predictive policing: Evaluating a Spatial-Temporal Forecasting Model in an operational deployment

The advent of big data has fuelled a machine-learning revolution, which in turn has included the emergence of predictive policing by law enforcement agencies, although little is known about the efficacy of this strategy and the conditions under which it can be effective in reducing levels of crime. The study involved an evaluation of a predictive policing pilot project implemented by the Vancouver Police Department in British Columbia, Canada, and focused on the use of a machine-learning system designed to conduct spatial-temporal crime forecasting on residential break and enters. It was also designed to contribute to the published literature on predictive policing and to offer guidance for police services considering adoption of this technology, by providing a template that could be used to assess the strategy’s effectiveness. The evaluation was conducted through the use of a pilot project that extended over 6 months from April to September 2016, during which time patrol resources were deployed to specific locations, at particular times of the day, as determined by a predictive policing model.
The effectiveness of the deployments in reducing residential break and enter rates throughout this time period was compared to a control period during which police patrols were not directed by the predictive model. A multimodal approach that included the use of geo-temporal data analysis techniques to evaluate the distribution, intensity of patterns, and volume of residential break and enters during the 6-month pilot project was compared to data obtained during the previous 4 years, to determine whether directed police resources based on the predictions, had an effect. The evaluation indicated the pilot project had a quantifiable effect in 4 out of the 6 months, with inconclusive findings for the remaining. The role of big data, machine learning, and artificial intelligence in forecasting, as well as its potential benefits and concerns were also discussed, with a focus on data integrity, ethics, and the human components of predictive policing.