By taking another look at the complex relationship between crime and society, researchers at the University of Chicago, including SFI External Professor James Evans, have developed an algorithm that can predict urban crime one week in advance with 90% accuracy.
The algorithm does this by detecting patterns in time and spatial coordinates of discrete events — rather than traditional neighborhood or political boundaries — from public data on violent and property crimes.
The model enables the team to examine how police respond to crime, demonstrating a disparate police response during crime spikes that led to more arrests in wealthier areas at the expense of disadvantaged neighborhoods, suggesting bias in police enforcement.
The study, published in Nature Human Behavior, analyzed eight cities — Chicago, Atlanta, Austin, Detroit, Los Angeles, Philadelphia, Portland, and San Francisco — with consistent results.
“We demonstrate the importance of discovering city-specific patterns for the prediction of reported crime, which generates a fresh view on neighborhoods in the city, allows us to ask novel questions, and lets us evaluate police action in new ways,” Evans said.
Read the paper, "Event-level prediction of urban crime reveals a signature of enforcement bias in US cities" in Nature Human Behavior (June 30, 2022)