I determined the optimal locations for 6 police stations in University City based on crime data from the last two years. I compared the results of three different approaches, all with the goal of minimizing the distance from the stations to crime locations for each occurrence of a crime.
Approach 1: Basic facility location
Approach 2: Require that each of the 6 geographic regions contains 1 facility
Approach 3: Limit the crime location capacity of each station
I used the intlinprog() function of Matlab to solve these integer programs. To save computational time and space, I first clustered any crime locations that were within 0.0009 decimal degrees, or about 0.05 miles, of each other. Instead of considering both of these locations, the program considers their average.
Use the menu on the right to see the details of each of these approaches.