GIS 4048 Module 5: DC Crime Mapping

Figure 1: Map of police stations categorized by number of crimes nearby; depicts the site of a proposed new police station.

Figure 2: Kernel density maps of three offence categories.

Our task this week was to make use of a variety of tools to analyze crime that occurred in Washington, D.C., in January of 2011. We began by geocoding police station locations using a CSV file. Only one of the stations needed to be manually located on the map. We then imported crime data from another CSV file. In order to analyze the crime and police station data, we performed two spatial joins. The first join involved a multiple ring buffer of half a mile, a mile, and two miles. This gave us an idea of the number of crimes that occurred at different distances from police stations. The second join was to police stations themselves, letting us know the relative number of crimes that occurred closest to each station.

The second layout depicts three kernel density maps, one for each of three crime categories. Kernel density maps work by using a user-provided radius (1500 square kilometers in the above case) and summing the values of each crime instance. The kernel density analysis assigns the highest value to the location of the incident; the value then decreases out to the radius. This gives us an idea of where potential crime hotspots may lie. Assuming the analysis is sound, it also helps mitigate the contingency of the exact locations of each crime in our interpretation. In other words, the specific location of a crime may be arbitrary, but the larger hotspots seen in a kernel density analysis may tell us more about areas of high crime potential and that need attention.