A crime prediction model based on spatial and temporal data

Abstract

In a world where data has become precious thanks to what we can do with it like forecasting the future, the fight against crime can also benefit from this technological trend. In this work, we propose a crime prediction model based on historical data that we prepare and transform into spatiotemporal data by crime type, for use in machine learning algorithms and then predict, with maximum accuracy, the risk of having crimes in a spatiotemporal point in the city. And in order to have a general model not related to a specific type of crime, we have described our risk by a vector of n values that represent the risks by type of crime.