Math models and policing

[Glenn E. Meyer]

This out of my expertise but found it interesting. So I wonder if folks in the know have an opinion. The idea is to use mathematical models to predict crime.

From the article, there are two factors:

1. The political aspect of the academics commenting on the usage. Not that really interested in that. Academic societies are becoming arbiters of members' social aspects and research nowadays. That has problems.

2. Does the methodology work? Anyone with experience? That's interesting.

https://www.popularmechanics.com/sci...tive-policing/

[Joe Mac]

My agency used "predictive policing" software for a while. Yep, it works -- and it tells you exactly what any 2-year-rookie officer working those beats could have told you. It's a huge waste of time & money.

[andre3k]

I think building this software based on old stats is useless. I guess its a modern spin on cop on the dot policing, which isn't new.

Now if they can apply this technology to scanning social media in real time and finding a way to disseminate that info quickly it would be much more useful. IG and Snapchat is where it's at for intel now.

Sent from my SM-G965U using Tapatalk

[ranger]

Data based policing will not work because the data (truth) will not fit the narrative.

[feudist]

50% of the time it works 100%.

[jnc36rcpd]

Since the now chief of my former department was in charge of the Synchronized Operations Command Center in a near-by, much larger agency, my former outfit has implemented predictive policing. Perusing the 2019 annual report, I find that a street in a bar district has the most disorderly calls rivaled by another street a couple blocks away. Another area has a bunch of domestics and some TFMV's while a fourth area, a congested residential and commercial community, (actually, to my surprise) has a bunch of traffic collisions.

Rent, dealing with the city, and covid closed some of the bars and restaurants so I guess those problems are solved. I don't know how you really deploy cops to prevent or interdict domestics and, good grief, the third "hot spot" is within walking distance of the Family Justice Center and the courthouses. I suspect the traffic issues are inevitable.

I'm not sure the department is getting all they pay the crime analyst for. Reading the annual report on predictive policing, you'd think he predicted the the Barrow/Parker gang was going to hit the Town Center Suntrust at 0920 Tuesday and be driving a black Ford coupe with 1931 Texas playes.

[KellyinAvon]

This out of my expertise but found it interesting. So I wonder if folks in the know have an opinion. The idea is to use mathematical models to predict crime.

From the article, there are two factors:

1. The political aspect of the academics commenting on the usage. Not that really interested in that. Academic societies are becoming arbiters of members' social aspects and research nowadays. That has problems.

2. Does the methodology work? Anyone with experience? That's interesting.

https://www.popularmechanics.com/sci...tive-policing/

So, "plan, do, study, act"? W. Edwards Deming would be pleased. As someone who has developed predictive equations to determine required manning based on workload factors? I would NEVER try to predict crime based on any kind of equation. Criminals are people and people screw up the numbers.

I'd go with an Intel-fusion/time-critical targeting approach.

Now don't get me wrong, if some woke asshole tried to tell me I couldn't do it because of some bullshit about the Police being the enemy? I'd do it anyway. It wouldn't work worth a shit, but I don't like being told how to think by idiots.

[Stephanie B]

How does any of that stuff beat tacking a large map to the wall and using colored push-pins to show where incidents have taken place?

[BehindBlueI's]

How does any of that stuff beat tacking a large map to the wall and using colored push-pins to show where incidents have taken place?

Modeling also shows times. I can have a bunch of catalytic converter theft push pins in a scrap yard, but day and middle shift don't need to be concerned if they are all happening at 0000-0400. I frequently get "patrol when possible" requests for my shift for problems that don't occur on my shift.

Modeling is also stupid, in the "garbage in, garbage out" meaning. I got my list of "hot spots" for violent crimes that officers are supposed to mark out at and be visible for 15 minutes a couple times a day. One is somewhere I *know* there is no violent crime. It's an area with a bunch of camp grounds, hiking trails, horseback riding places, etc. It's also the home to a bunch of unofficial, but not illegal, ranges. Neighbors call in "shots fired" runs all the time because of guys using those ranges.

As a sergeant, the first thing I ask after reading these things at roll call is to the guys that work that area if it makes sense and what additional info they have just because I know the automated data results in that sort of thing on occasion.

[perlslacker]

I don't know much about policing, but I have a short handful of computer science degrees and I used to work for a company whose product was heavily AI-based.

When machine learning algorithms are fed good data, and the learning process is designed by folks who know what they're doing, you can achieve great results. To answer a question that was already asked, it will be better than a map w/ pins because in-depth statistical analysis will frequently catch links that humans miss. The analysis will also be faster, and free up humans to do other things.

When machine learning algorithms are NOT fed good data, or the learning process is NOT well-designed, then you can run into trouble. Machine learning is a sexy sales buzzword right now so there are a lot of hucksters who don't know what they're doing looking to cash in.

So it'll be interesting to see if this works out. I'm not 100% sure how I feel about the morality of this kind of predictive policing yet.