This post is part of the STEM Story Project series.
What if an equation could put police officers at the scene of a crime, before it happened? In Southern California, a team of academics discovered that crime patterns could be mathematically modeled like the way that earthquakes and their aftershocks are modeled. The LAPD gave this “predictive policing” a chance to prove itself on the streets, and discovered that it worked. Now, this mathematical model could transform the future of law enforcement.
Producer Aaron Mendelson first came across this idea through an academic paper that aspired to refine a statistical model that predicts crimes. This piqued his interest, and he began to dig deeper into the world of crime modeling. What he found was that it was not only something being explored in academia, but it was also in use by one of the nation’s largest police forces, the Los Angeles Police Department. The software is called PredPol, which takes crime data, runs it through an algorithm, and visualizes it on maps.
“The epicenter of this work is in LA — both at UCLA’s MASC (Mathematical and Simulation Modeling of Crime) Project and at the LAPD,” says Mendelson. “So I hopped on a plane to Southern California to figure out what was going on down there.” In reality, Mendelson’s journey to collect audio for this story involved much more travel than a simple plane ride to SoCal. His trajectory included stops in Milwaukee, where he had just moved for the summer; Los Angeles, where he interviewed people at the LAPD and UCLA; Washington, D.C. and rural Pennsylvania, where he covered another story; and then Rockford, Ill., to attend his cousin’s wedding. Mendelson sums up his trip in just five words: “It was a hectic week.”
In Santa Clara, CA, Mendelson was able to talk to George Mohler, a mathematician and one of the leading postdocs working on this model. What he found was an unexpected connection between earthquakes patterns and crime patterns:
“Mathematically, earthquakes and crime work in a similar way. Mathematical models for predicting earthquake aftershocks could be applied to predict the ‘after-crimes’ of an initial incident…According to Mohler’s model, one crime sets off a wave of crimes in an area. The equation draws in details from police reports, such as times, locations and types of crimes that already have happened.”
Mendelson’s piece ran on The California Report in Sept., where they also noted that San Francisco plans to implement the system by the end of 2014 and hopes to add other capabilities, including predicting gun violence and property crimes. Take a listen to his piece — we predict that it’ll teach you something new.