Module 1: Crime Analysis

Crime Analysis

Welcome back blog readers! It's the first week of GIS 4048 - Applications in GIS.

We kicked off the course with hotspot analysis exploring grid overlay, kernel density, and Local Moran's I.

Part of this week for me was getting reacquainted with the features and geoprocessing tools in ArcGIS Pro. I had not used some of them since our Cartography course. 


Analysis

Grid Overlay

I had a small issue with this on the Spatial Join. I was working through all the options while waiting on a reply in the discussion forum. I was a few tries away from the correct settings when my classmate replied. 

Other than that, grid overlay analysis was straightforward. Brushed up on using the export feature with SQL filters, the dissolve tool, adding a field, and calculate field. 



Kernel Density

First time using the Kernel Density tool (well the 2nd time if you count the first part of the assignment). I used the parameters in the assignment and didn't have any issues. Also the first time using Reclassify and Raster to Polygon tools- but between the standardization across geoprocessing tools and the assignment instructions I had no issues. 



Local Moran's I

Also a first time use of the Local Moran's I tool. Again by following the assignment and using the skills from the previous steps, I did not have any issues. 



Results

There were 4 categories to compare between the 3 hotspot maps. 
  1.     Total area (mi2 ) of hotspots in 2017
  2.     Number of 2018 homicides within the 2017 hotspots
  3.     Percentage of all 2018 homicides within the 2017 hotspots
  4.     Crime density (2018 homicides within the 2017 hotspots per mi2 )

Grid Overlay 
  • Highest crime density but only by a small percentage. 

Kernel Density

  •     Largest total area     
  •     Most number of 2018 homicides within 2017
  •     Highest percentage of 2018 homicides within the 2017 hotspots. 

Local Moran's I 

  • Lowest numbers in all 4 categories. 

If I had to brief Chicago's Police Superintendent on where to allocate limited policing resources, I believe I would use the Moran’s High-High cluster map. Overlayed on the district/beat map– it would highlight the "hottest" areas and the respective districts can deploy more officers to those "hot" beats. 

If this was a regular Deployment meeting I could make an argument for Kernel or Grid based on the needs of the department. 

For a more "on-demand" allocation of resources – there are GIS data analysis programs out there that can use prior years data and generate dynamic maps that can suggest which area the next “call” might be at any given day and time frame so resources can be adjusted accordingly. 


Notes

I thoroughly enjoyed learning the skills associated with Crime Analysis. I have a future work project planned that will use these particular skills. Being able to map "hotspots" more effectively will hopefully reduce response times. 


And since we're talking about Crime - RIP Capt. Raymond Holt (Andre Braugher). 




Comments

Post a Comment