Experts from the University of Chicago have created an algorithm that predicts which crime will be committed after 1 week. The accuracy level of the forecast given by the created algorithm is 90%. However, such accuracy is possible only in predictions made by the algorithm on a small area. Currently, the mentioned algorithm is considered one of the leading algorithms in the world. This system is based on a trained neural network. That neural network analyzes the regularities of places, times and crimes themselves.
This entire process is based on open source data on violent and property crime. Experts have already tested this algorithm in Chicago, Atlanta, Austin, Los Angeles and other big cities. The system of the mentioned algorithm works like this: This or that city is divided into square parts with a size of 300 meters. Later, the system analyzes the times and places of individual crimes and determines certain regularities, and as a result, predicts future criminal events.
"We have created a digital version of the urban environment. If the system is given information about past events, it can predict future events. This is not a wave. There are limitations in this matter. But we have tested this system and it works very well. Now it can be used as a modeling system. In this case, one can see what will happen if the number of crimes increases in one of the city districts or if police control is strengthened in another district. If all these variables are used, it will be possible to see how these systems evolve." - said Ishanu Chattopadhyay, an associate professor at the University of Chicago School of Medicine.