Lab 6 - Crime Analysis

This week we focused on a valuable and well-discussed topic: crime analysis. While crime data analysis has been around since the 1800s at least, there is new focus on spatial scale, geography of crime events, temporal considerations with crime, and GIS mapping capabilities. Using modern GIS technology analysts and criminologists can utilize a variety of tools and methods to analyze crime, a few of which we used in our lab.

The lab started with some basic analysis involving aggregating police crime data to determine crime rates, creating two choropleth maps, and a scatterplot graph to look for potential correlations between residential burglaries and types of home ownership. From there, I moved on to hotspot identification, with a hotspot being an area that has a greater than average number of criminal events, or a greater number of victimizations. To find these hotspots I again used police data to separate out specific types of crime for further analysis, including auto theft and burglaries. I then performed a few different analysis methods for practice and comparison.

Grid-based thematic mapping determines the number of crime events (2007 burglaries in Alburqueque, NM in this case) per grid cell laid over a study area. Using spatial joins, SQL queries, classification thresholds, and the dissolve tool I created a series of polygons in a study area reflecting areas of high burglary concentration. I then created a kernel density map with the original data, looking at density areas with at least 3 times the mean, which I converted to polygons. Finally, I created a Local Moran's I map, again doing spatial joins and SQL queries, to determine clusters with High-High values. After calculating the total area for each method's hotspots I used crime data from 2008 to compare the potential accuracy and predictive capabilities of each method. Hotspot area, crime density of 2008 burglaries overlaid with 2007 burglaries, and the percentage of crime in the hotspots versus overall crime for the area were all considered. Each method had their advantages and disadvantages when considering resource allocation, some producing larger areas which captured more of the total crime in the city (grid-based) and some with smaller areas but a higher density of crime found in each. Below is the final map comparing all three methods.

Crime analysis with GIS was interesting. It is a growing field, with a variety of methods and applications being studied and discussed, definitely lots of room to grow for someone interested in learning more.

Crime hotspot map comparing three methods of analysis: Grid-based Thematic, Kernel Density, and Local Moran's I.