A Crime Early Warning System: Using Spatial Statistics to Detect Changing Geographic Patterns in Crime

Large quantities of spatial data can be as much a burden as a boon without the tools to properly tease out important information. For police departments, HunchLab enables early detection of changes in patterns of crime, pulls information automatically out of millions of incident records, and provides the means of detecting and stopping spikes in crime earlier than they would be found through more conventional means.

The cost of crime is not only measured in terms of injury, loss of life and property, law enforcement, court and corrections, but it also imposes substantial indirect and social costs through insurance rates, blight, damage to education, community and family disintegration, damage to the tax base and property value. Criminal activity is an incredibly dynamic phenomenon. Patterns shift and change daily and weekly in response to changes in target behavior, police intervention, business activity and other factors. Increasingly, the police commander, the special unit and the officer in the street have access to technology that makes their professional work both safer and more effective, but street crime remains a perniciously difficult phenomenon. From electronic maps to databases, police staff have greater access to information about the incidence of crime, but the sheer volume of data in an urban police department is crushing.

Now imagine you are a police captain, and you could receive an e-mail on your Blackberry every morning summarizing how the district’s crime patterns changed in the past 24 hours. Or imagine that an officer responding to a domestic disturbance is able to place an alert on a particular address so he will be notified if any other calls occur at that address – if these additional domestic disturbance calls occur, they may be a precursor or signal of domestic abuse for which intervention by social services could prevent a aggravated assault or even a homicide. HunchLab is a software tool developed for law enforcement organizations to enable these features. It is essentially an ‘early warning system’ for events that a) have specific geographic locations, b) occur in relatively large numbers and c) change in frequency over time.