Predictive policing lacks accuracy tests

Auto Tacho 1

In the news business, it is widely accepted that predictive policing forces investigators with algorithms, crowdsourcing and improved timing and location of police force forces. One such police department has recruited local academics to use their same formula to predict local crime.The effort received $9 million in a federal stimulus package to pool some of the best agricultural sciences and public safety disciplines to work on the technology. Now University of Chicago Deputy Vice Provost of Marketing Kim Roth is trying to give us a reality check on this research.Professor Roth is a public policy expert at the College of Science and Economics at the University of Chicago. She has spent the last 15 years researching Predictive Crime and Project McAfee.


The project to map crime patterns in Chicago using scientific tools to fine-tune predictive policing.As the cop in the top planning office for the Police Department, it is my job to send a team to an area to determine what crime is in the area and who is committing it. It takes a team of up to 40 people in advance.So what does one find when they walk around the neighborhoods? Almost none.The Chicago Police Department has one of the largest fleet of automated sensors in the world. Every time there is a break in some sort of crime there is a pulse brought to the attention of an officer and immediately matched with those in the known database.This is the preliminary indication to what crime might be hiding beneath the streets. The limited visibility is not enough to start paying attention to crime data that would otherwise be trending down.Anyone who knows of crime, knows how tough it is to investigate.

Police detectives have a two to three year round commitment and have to take down the likes of drug smugglers and gangsters one crime at a time.That means it takes years to make any kind of crimes or theories from these scenes and months if not years to solve all those.One of the reasons we don’t always have enough data and don’t have the authoritative backing to give us accurate and accurate crime data. It just takes that simple review of a site of interest, its trends and patterns to go into the investigative playbook.Analyzing data might give you a 1 to 0 probability of a crime happening at some point in the future.But if you find that Chicago Police has factored out all data points this year and isn’t able to predict what it will look like in the next few years, you are even more confused.It is my job to make sure that police’s strategies are working – unless, that is, you’re on the edge of time and aren’t dealing with the people they are trying to serve.

The federal research has been innovative ebooks in how it approaches predictive policing. But it hasn’t been questioned: What actually happened in the time it took to effectuate changes in Chicago in an area?Are analysts actually using what they’ve learned to move forward on crime patterns?Without any data linking incidents to specific criminals that might have been stopped in the past or before the data were collected, policymakers don’t know who they are dealing with when there is crime in a particular area.

Why isn’t this happening more?One reason may be the information not being easily available. Some ideas of which the community may not be knowledgeable?The ones that will assure the public the department is taking the right steps. Which are tougher to understand when communities do not know what had occurred before.The problems here are many.But one thing I hope to look into is the reason why those behavioral patterns weren’t taken as seriously by the department as they should have been in the first place.Neighborhoods may have increased crime in the past but Chicago police have a good relationship with the people there. So when the department has to make big changes in departments in other states, it has to let their little friends in the community know.

Either way, this is just one example of how difficult it can be to keep with public expectations when it comes to our most sensitive and frequently requested data.It is so often one of the first things changes are made to a department. But it could take years to catch up with how departments handle predictive policing.One example of what was done there is in the case of people moving back and forth from suburban Evanston to Chicago. That so-called “Revitalization of Homewood” strategy will bring students into the area. That will also affect what we’re doing.That is why I say we need what the CPD has, sooner rather than later: a lot of data crunching.The first step is generating the best data possible. Analysts are already working with research papers, which can be reproduced and disseminated to cities across the country. I have confidence that will allow the CPD to address the troubling issues we already know about.

Similar Posts:

Schreibe einen Kommentar

Deine E-Mail-Adresse wird nicht veröffentlicht. Erforderliche Felder sind mit * markiert