GIS Applications: Chicago Homicide Crime Analysis Hotspot Techniques

This week explored how different spatial analysis techniques can identify homicide hotspots in Chicago, using 2017 data to evaluate how well they predict 2018 incidents. Three methods were applied: grid overlay, kernel density estimation, and Local Moran’s I. Each of these methods offer a unique view of crime area in the city.  

The grid overlay method divided Chicago into half-mile squares and selected the top 20% with the highest homicide counts. This approach was simple and direct, producing clean polygons ideal for operational planning (Figure 1).

Figure 1

The kernel density technique produced a surface showing where homicides were most concentrated, based on how closely the incidents were located to each other. This method helps highlight areas with higher crime intensity without breaking the map into fixed units, giving a clearer view of crime clusters (Figure 2).

Figure 2

The Local Moran’s I method identified clusters of high homicide rates by finding areas where high rates were near other high rates. Although it covered the largest area, it also captured the most 2018 homicides (Figure 3).

Figure 3

I looked at each map to see how well it predicted the 2018 homicides by counting how many incidents fell inside each hotspot and checking how big those zones were. Local Moran’s I caught the most homicides, but its hotspots covered a lot of ground and weren’t very precise. The grid and kernel density methods showed smaller, tighter hotspots that would be easier to focus on for real-world action.

The lab showed there can be a trade-off between the size of a hotspot and how accurately it predicts crime. Bigger zones catch more incidents but can be too broad to really target resources. Smaller hotspots might miss some crimes but give clearer, more practical areas to work with. In the end, this lab clarified how different mapping techniques can actually help shape crime prevention strategies in the real world.