The news story of the assault down the street or the robbery on the corner often makes us think about the risk of becoming a victim of such an event.  But why do events happen where they do? One way to explain the spatial pattern of crime is to contextualize locations using other data sets.  Is there a bar on the corner emitting boisterous patrons?  Are there a lot of bus routes heading through the area providing an easy means of escape for the criminal?  These geographic risk factors can help police to model crime and explain why crimes happen where they do (or at least describe what geographic features correlate with crime).

Rutgers Center for Public Security has released the RTMDx utility that Azavea developed earlier this year.  This new desktop software tool automates one type of crime forecasting using contextual geographic risk factors such as bars, bus stops, and schools.  The Rutgers team has done a great job of encouraging police departments to think not only about where crime is occurring but what is located near these hotspots of activity through their Risk Terrain Modeling methodology.  This environmental criminology approach to analyzing crime does much to help law enforcement agencies to focus on the potential causes of crime and not just its manifestation.  But why did we need to build a utility?

Forecasting crime risk requires a good grasp of statistics and getting those statistics right is not always easy. A small mistake along the way can easily invalidate your results. The Risk Terrain Modeling methodology uses raster (a grid of evenly spaced cells) data sets to represent the different factors.  But a raster data structure precludes using the traditional linear regression models available within conventional desktop GIS software such as ArcGIS.  In collaboration with William Huber, I developed a new statistical approach to Risk Terrain Modeling that solves many of the challenges faced in this arena. This new approach forms the basis of the RTMDx utility that Rutgers released. It also makes this type of predictive analysis available to analysts without proprietary statistical software.  Under the hood, the utility uses the free and open source R framework to conduct statistical modeling.  This development approach increases transparency by enabling anyone with a copy of the utility to examine the choices made within the utility and to suggest improvements.

We’re glad that we could support Rutgers’ effort to advance the work of police departments across the country.  A copy of the utility is available for free download, and the documentation, including some additional details on the statistics, are also available.  Rutgers is also offering training seminars on using the software.  Lots of resources; check it out.