This is my work on using a p-median model to redistrict patrol areas in Carrollton, TX. In particular, this incorporates workload inequality constraints into the p-median model.
The final published paper is:
Wheeler, Andrew P. (2019) Creating optimal patrol areas using the P-median model. Policing: An International Journal 42(3): 318-333.
You can also access a pre-print version here or on SSRN.
The Jupyter notebook shows how to use the python pulp library to set up the linear programming problem. And ArcGIS and the network extension to get the travel time distances and weights needed to optimize areas.