In this week's module, we learned different mapping techniques to analyze crime hotspots. The images of the maps above show three different analyses of crime hotspots in Chicago, IL.
We created a grid-based thematic map to find the highest number of homicides per 1/2-mile Chicago grid square. I used the Spatial Join and Select by Attributes tools to create a new feature class that contained the number of homicides per grid cell and removed the grid cells where there were no homicides. I selected the top 20% of the grid cells with the highest count and used the Dissolve geoprocessing tool to make the feature class one polygon. The result is a thematic map that shows the areas where the top 20% of homicides occurred in Chicago in 2017.
To create a kernel density map, we used the respectively named geoprocessing tool to set a search radius of a 1/2-mile distance. I edited the symbology on the map based on the mean value of 5.63. I used two classes: below 3 times the mean, and above three times the mean. This symbology setting showed me the density area I needed, but I had to narrow down my data further. I used the reclassify tool to convert the raster class that was 3 times the mean into a polygon. The result is a kernel density map that shows the areas with the highest clustering of homicide points in Chicago in 2017.
We created a Local Moran’s I map to show the rate of homicides per household. I used the spatial join tool to find the number of homicides within each Chicago census tract, and then calculated the number of homicides per 1000 housing units. I used the Cluster and Outlier Analysis (Anselin Local Moran’s I) spatial statistics geoprocessing tool to find clusters of high crime density. This tool interpreted the different kinds of cluster values, and for this crime map we used the high-high (HH) clusters. These HH clusters show where homicide rates are high relative to the proximity of other areas in the city with a high homicide rate. I selected the 130 HH clusters from the results to create a new feature class, and I used the dissolve tool to create a new polygon. The result is a Local Moran’s I map the shows where the highest rates of homicide are per household relative to the rates of other areas with a high homicide rate.
The kernel density map is the best map for predicting future crime hotspots. I think that the kernel density map clearly shows the areas where homicide rates are most highly clustered. With a density of 196 2018 homicides within the 2017 hotpots per square mile, the kernel density map successfully predicted the area of the highest crime rate for 2018.
Comparatively, the Local Moran’s I map predicted the highest amount of crime in the largest area per square mile. I think that the Local Moran’s I map gives the best range of area, but I don’t think that it clearly identified the specific hotspots within the city
I believe the kernel density map would be most useful to the City of Chicago’s police chief. Of the three maps, the kernel density map shows the smallest total area, with the highest crime density. The map identifies 3 square miles where the City should allocate their policing resources the most.
Very interesting Emily! Great job!
ReplyDelete