More

    AI Algorithm Can Predicts the Future Crimes

    A new computer model uses publicly available data to accurately predict crime in eight US cities, while showing an increased police response in affluent neighborhoods at the expense of less favored areas. Advances in artificial intelligence and machine learning have  sparked the interest of governments wanting to use these predictive policing tools to deter crime. However, early efforts to predict crime have been controversial because they fail to account for systemic biases in policing and its complex relationship to crime and society.

    Data and social scientists at the University of Chicago have developed a new algorithm that predicts crime by learning temporal and geographic patterns from public data on violent crime and property crime. It has proven successful in predicting future crimes a week in advance with about 90% accuracy. In a separate model,  the research  team also examined police response to crime by analyzing the number of arrests after incidents and comparing these rates between different neighborhoods and socioeconomic status.

    They saw that crime in affluent areas led to more arrests, while arrests   in  deprived neighborhoods fell. Crime in poor neighborhoods, however, did not result in more arrests, suggesting bias in police response and enforcement. When you stress the system, it requires more resources to arrest more people in response to crime in a wealthy area and draws police resources away from areas of lower socioeconomic status, said Ishanu Chattopadhyay, PhD, an assistant professor of medicine at the UChicago and senior author of the new study, published June 30, 2022 in the journal Nature Human Behavior.

    The new tool was tested and validated using historical data from the City of Chicago on two broad categories of reported events: violent crime (homicide, assault, and batteries) and property crime (burglary, theft, and auto theft). This data was used because it was most likely to be reported to police in urban areas, which have a history of distrust and lack of cooperation with law enforcement. Such crimes are also less susceptible to enforcement bias, as is the case with drug-related crimes, traffic delays, and other misdemeanors.

    Previous efforts to predict crime often use an epidemic or seismic approach, depicting crime as emerging from hotspots that spread to surrounding areas. However, these tools neglect the complex social environment of cities and do not consider the relationship between crime and police impact. Spatial models ignore the city’s natural topology, said sociologist and co-author James Evans, PhD, Max Palevsky Professor at UChicago and the Santa Fe Institute.

    Transport networks respect roads, sidewalks, train and bus routes. Communication networks respect areas with a similar socio-economic background. Our model enables the discovery of these connections. The new model isolates crimes by looking at the temporal and spatial coordinates of individual events and recognizing patterns to predict future events.

    It divides the city into spatial tiles about 1,000 feet in diameter and forecasts crime in those areas rather than relying on traditional neighborhood or political boundaries, which are also biased. The model performed equally well with data from seven other US cities: Atlanta, Austin, Detroit, Los Angeles, Philadelphia, Portland, and San Francisco.

    “We demonstrate the importance of discovering city-specific patterns in predicting reported crime, which creates a new perspective on city neighborhoods, allows us to ask new questions and evaluate policing in new ways,” Evans said . 

    Please note that the tool’s accuracy does not mean it should be used to guide law enforcement, as police use it to proactively raid neighborhoods to prevent crime. Instead, it should be added to a toolbox of city policies and police strategies for fighting crime. We created a digital twin of urban environments. If you feed it data from the past, it will tell you what will happen in the future. It’s not magic, there are limitations, but we’ve validated it and it works really well, Chattopadhyay said. Now you can use this as a simulation tool to see what happens when crime increases in one area of ​​the  city  or  enforcement increases in another. If you apply all of these different variables, you can see how the systems evolve in response.

    The Cents Warrior – Please do subscribe.

    Recent Articles

    Related Stories

    1 Comment

    Leave A Reply

    Please enter your comment!
    Please enter your name here

    Stay updated - Get the daily news in your inbox