Knoxville Police hope data project could predict, prevent issues

Sept. 12, 2016: Knoxville police are using a new data program to identify officers who are at risk of having an adverse incidents in the future to help prevent those incidents.

Knoxville police are working with nationally renowned data scientists in an effort to keep their officers, and the citizens they serve, safer.

For more than a year, the department has been providing raw data about their officers and calls to researchers at the University of Chicago’s Center for Data Science and Public Policy.

Data scientist Joe Walsh is a member of the team.

“When you have a big haystack, it’s hard to find a needle,” Walsh said.

But Walsh and his team believe they can use data from law enforcement agencies to predict what they call ‘adverse incidents,’ such as officer involved shootings or excessive use of force. They also consider other factors, like the race and ethnicity of officers and those they interact with, training, geography and past jobs.

By cross-referencing all these data points, they believe they can head-off an officer’s potential issue.

“So the police department can provide them with the training and support they need to prevent those things from happening in the first place,” Walsh told WBIR 10News by Skype.

Still, he noted it’s a tough problem to crack, since officers perform a difficult, complicated and often dangerous job.

The researchers began by partnering with Charlotte, N.C. police before expanding to Nashville PD. Last year, Knoxville joined the study. They now supply the Chicago team with information about calls for service, citizen complaints and internal affairs data, among other things.

The project was sparked as part of a White House initiative to improve police in America.

Top officials in KPD hope it will allow them to nip potential problems in the bud. Using this ‘big data,’ the researchers believe they can identify an officer experiencing abnormal stress – someone who might be more likely to make a mistake in the field.

“The whole goal for me is to identify an officer before they start exhibiting those behaviors, or as soon as they begin to exhibit those behaviors, so we can pull them back in and save their career,” said KPD Deputy Chief Gary Holliday.

Holliday said the department has had an existing Early Intervention System (EIS) for some time now, but it was more rudimentary than this new project and was based on accrued points totals that then flag an officer for supervisor review.

“Unfortunately the existing EIS systems in use around the country are not very good,” Walsh said. “They flag a lot of officers who do not go on to have adverse incidents while missing a lot of officers who do.”

Holliday said he’s excited by the possibilities of a more comprehensive system.

“One of the things I think is important, if we can get to the point where we can identify some character traits of certain behaviors you can screen out in the hiring process,” Holliday said. “To me, that’s something that would really be a valuable tool for law enforcement, actually any kind of employer.”

Walsh said feedback from officers has been positive. Already, they’ve learned what calls make the job more difficult for officers – suicides and domestic calls involving young children. He said in interviews, officers told the team those calls add to their stress level more than others.

He also noted that such a system is not meant to be a quick-fix – any results the computer returns would need to be reviewed by a supervisor – a real person, not an algorithm.

Still – he says using big data like this has the potential to solve a big problem.

“It seems like something that data can be used to improve what’s out there,” Walsh said. “If that’s the case, and they can prevent bad things from happening, we’re certainly interested in helping.”

“It’s just another tool in the toolbox, but it’s a handy tool,” Holliday said.

Researchers are still gathering data from KPD, and have not implemented the new EIS system yet, but they hope to in the coming year.

(© 2016 WBIR)


To find out more about Facebook commenting please read the
Conversation Guidelines and FAQs

Leave a Comment