Applications of GIS: Crime Analysis

Hello Everyone! 

This week I've been enjoying a much-needed spring break, and have been focusing on potentially testing out of one of my future classes. For this project, I was assigned a discussion on the assumptions that hot spot maps can make and how that can influence crime analysis decisions and I was also tasked with a lab that had me create several types of crime analysis maps using ArcMap 10.6. While the first two portions of the lab had me create choropleth and hotspot maps the portion I would like to focus on is the last part using three various forms of hot spot maps to predict future crime trends. The three main methods can be seen below and I will briefly discuss them in depth, as well as discuss which I think is the most suitable.

 The map above compares the three types of crime analysis hotspot maps in question. The first is a grid-based thematic map. To create this I took burglaries from the year 2007 and spatially joined them together with a grid of the area. I then selected the top 20% of grids that had the highest counts of burglary crimes in each after excluding cells with a count of '0'. I then created a polygon of this grid that shows where the highest burglary activity is located throughout the region. Next, I created a Kernel Density map. To do this I took and created a density map of the burglaries from 2007 and found where the density was highest, this gave me a completely different result as I used a half-mile search radius to help shape my map. Finally, I made a Local Moran's 1 map that uses cluster and outlier analysis to find areas of high concentrations of crime near other areas of high concentrations of crime. I then stacked these three map types to get a comparison and find which method would best suit crime forecasting.

After looking into the numbers and analyzing the data I found that if I were to be a police chief or a sheriff, I would want to use a Local Moran's map. This is because it is neither too big (like the grid) or too small (like the kernel density) to focus the efforts and resources on. I also notice that the areas of high concentration in the Moran's map are pretty central to my data and share the same geographic location with the areas of crime with the other two analysis maps. When looking at the numbers, I found that when I compared the 2008 burglary data to the data from 2007, the Local Moran's map did the best at forecasting for the next year.

I have been fortunate to have a background in Crime Analysis through an amazing opportunity to be an intern analyst at the Jacksonville Sheriff's Office. Crime analysis is one of my desired fields of interest and creating maps like this has always been one of the things I love most about GIS. It has been an absolute pleasure pursuing this project, and as always, ~ Map On