Track: Emergency Management and Public Safety
Smoothing Crime Incident Data: New Methods for Determining the Bandwidth
Doug WilliamsonCrime analysts are increasingly using GIS to analyze and display geographic concentrations or "hot spots" of crime events. One of the methods for doing this is kernel smoothing, a spatial statistical method that generates a map of density values from point event data. A critical issue in kernel smoothing is the selection of the bandwidth size - the radius of the circular area in which smoothing is performed. Most GIS and spatial statistical programs that perform kernel smoothing calculate the bandwidth based on the size of the study area. These estimates of bandwidth do not reflect the geographic distribution of points within the study area, only the area's geographic extent. This can result in misleading density values and maps that are either too smooth or too spiky in appearance. We propose a new approach for bandwidth estimation based on nearest neighbor distances among points. This method offers an improvement over existing methods because it is based on the spatial relationships among the points. We illustrate the use of this method in creating maps of crime density utilizing multiple bandwidths in Brooklyn, New York.
Doug Williamson
Center for Applied Studies of the Environment (CAPSE)
(10th Floor Hunter North)
695 Park Avenue
New York, NY 10021
USA
Telephone: (212) 650-3591
Fax: (212) 650-3528
E-mail: dougwill@everest.hunter.cuny.edu