Model of burglary patrol allocation
I was working at LSU and a friend of mine was a local cop. He had been a homicide detective but got involved in some union activity so was being punished for that by being re-assigned to the Research and Planning Department. He asked me for some help with a research project he was doing in support of a burglary task force. I got a couple of people from LSU to help. It resulted in a publication, Model of burglary patrol allocation, Journal of Police Science and Administration, 5(2) (June 1977). pp 179-184 by Phil Boudreauz, Gary Carson, John Pisa, and Chris Schroeder.
We did a spatial forecasting model, where we forecast the probability of a commercial burglary in the next week at a block-face level -- i.e, the odd street number side of the 700 block of XYZ Street.
The idea was to then rank locations by those probability estimates and conduct surveillance. It didn't really work out as planned. Not because the model didn't work well, it actually did. But because cops think surveillance is boring and they won't do it if they can figure out a way to get out of it.
We should have realized that after our first discussion with the sergeant who ran the task force. Pretty much all he wanted to talk about was the cool cars they got to drive as part of their undercover personas. This was way before forfeitures, which is the way cops get cool cars to drive today. Back then they went to car dealers and extorted loaners from them by implying that if they didn't donate a car they'd be left without police protection.
Then they'd drive around in the cool cars to impress potential informants. (and street hookers, etc, etc).
The reason for our research project was that the cool cars and informant approach wasn't working, they weren't making any burglary arrests. The second day that our forecasting model was in operation there was a burglary at the top ranked prediction location.
They didn't make an arrest then either because they weren't using the model to do surveillance. They just didn't want to do surveillance. It was boring.