Seattle Mayor Mike McGinn announces the expansion of the city's data-driven policing effort. Photo courtesy of City of Seattle.
Seattle Mayor Mike McGinn announces the expansion of the city’s data-driven policing effort. Photo courtesy of City of Seattle.

Back in February, Seattle rolled out a new “Predictive Policing” software in the East and Southwest Precincts, hoping to keep the city safe by using data that better predicts where crime may happen.

Now, after “anecdotal success” with the software so far, it’s being expanded across all five precincts in the city, Mayor Mike McGinn announced yesterday at a press conference.

A "Predictive Policing" software screengrab from Elgin, just northwest of Chicago.
A “Predictive Policing” software screengrab from Elgin, just northwest of Chicago.

But for the software to be most beneficial, McGinn wants citizens to be proactive about reporting crimes.

“We’re asking the community to get involved by reporting even minor property crimes so we can improve our data set and predict where crime is likely to occur,” he said in a press release. “This is a tool that can help us prevent some crimes before they happen, so it’s very important that community members get involved.”

The software inputs historical crime data from 2008 into an algorithm that predicts where and when crime is likely to occur to a geographic area as small as 500 x 500 feet. It’s estimated to be twice as effective as a human data analyst working from the same information.

Officers are provided with the forecasts before each watch shift. They will know specific locations within their sectors where the data shows crime is more likely to occur based on past history. Officers use 70 percent of their time responding to calls and the other 30 percent patrolling.

“The Predictive Policing will help us figure out how to intelligently using the 30 percent proactive time,” McGinn said in February.

Built on the same model for predicting aftershocks following an earthquake, the software was designed at UCLA and piloted by the Los Angeles Police Department for one year, reducing crime by 13 percent. It’s being used by other police departments around the nation.

“Predictive Policing” does not use names or any other identifying data — just date, time, location and type of crime.

“This will help remove bias from the equation,” McGinn said. “The software helps take that out.”

The software will cost the city $73,000 with a $45,000 subscription fee every year. This is all part of McGinn’s SPD “20/20 Vision for the Future” under Initiative 14, which encourages using data-driven practices for deployment.

Like what you're reading? Subscribe to GeekWire's free newsletter to catch every headline


  • Anonymous

    Since it’s based of historical data, we could assume that this software is suicidal. As the software & police get better to predict and prevent crime, less data become available, impacting its own accuracy.

  • Waterside

    This sounds suspiciously like red-lining. It would be hugely controversial in a city with a large minority population.

Job Listings on GeekWork

Find more jobs on GeekWork. Employers, post a job here.
Creative DirectorUIEvolution Inc.
Sr. iOS DeveloperUIEvolution Inc.