Homicide prediction:

In this session of the Applied Mathematics Seminar we present the main results obtained in the application of GLoG filter to homicide data from 2013 to 2019 in Bogota. For this purpose, a graph was constructed where the nodes correspond to the police quadrants of the city and the edges link the quadrants that are adjacent. The GLoG filter was applied to this graph and both dynamic and static features were obtained, with which a logistic regression model was trained. Among the main results, it was found that if the model assigns 10% of the quadrants as Hotspots, it correctly predicts 33.8% of the homicides.