Scientists at the University of Chicago have developed a new algorithm that can predict future crimes a week in advance with 90 percent accuracy and a range of about 1,000 feet. It makes predictions by learning patterns from public data on violence and property crime. The tool was tested and validated using two types of reported incidents by the City of Chicago: violent crime (homicide, assault, and battery) and property crime (burglary, theft, and motor vehicle theft). Both types of data are used because they are the most likely to be reported to the police in urban areas that have historically lacked trust and cooperation with law enforcement. Unlike drug offenses, traffic stops and other misdemeanors, both types of crimes are also less prone to law enforcement bias. The new model divides crime by examining the temporal and spatial coordinates of discrete cases, detecting patterns to predict future crimes. It divides the city into “spatial slices” about 1,000 feet wide and predicts crime within the area. Previous models relied more on traditional neighborhoods or political boundaries, which were subject to bias. The model performed equally well on data for seven other U.S. cities: Atlanta, Austin, Detroit, Los Angeles, Philadelphia, Portland, and San Francisco. Lead author Ishanu Chattopadhyay is careful to point out that the tool’s accuracy doesn’t mean it should be used to guide law enforcement policy; Chattopadhyay said, for example, that police departments shouldn’t use it to proactively gather in communities to prevent crime. Instead it should be added to the toolbox of urban policy and policing strategies to tackle crime. “We create a digital twin of the urban environment. If you feed it data that has happened in the past, it will tell you what will happen in the future,” he said. Chattopadhyay added: “It’s not magic; there are limitations, but we It’s been validated and it works really well.” “Now you can use it as a simulation tool to see what would happen if crime increased in one area of a city, or increased law enforcement in another. If you apply With all these different variables, you see how the response of the system evolves.” The research paper was published in the journal Nature Human Behavior.
This article is reproduced from: https://www.solidot.org/story?sid=72000
This site is for inclusion only, and the copyright belongs to the original author.