Multi-density crime predictor: an approach to forecast criminal activities in multi-density crime hotspots

Abstract

AbstractThe increasing pervasiveness of ICT technologies and sensor infrastructures is enabling police departments to gather and store increasing volumes of spatio-temporal crime data. This offers the opportunity to apply data analytics methodologies to extract useful crime predictive models, which can effectively detect spatial and temporal patterns of crime events, and can support police departments in implementing more effective strategies for crime prevention. The detection of crime hotspots from geo-referenced data is a crucial aspect of discovering effective predictive models and implementing efficient crime prevention decisions. In particular, since metropolitan cities are heavily characterized by variable spatial densities of crime events, multi-density clustering seems to be more effective than classic techniques for discovering crime hotspots. This paper presents the design and implementation of MD-CrimePredictor (Multi- Density Crime Predictor), an approach based on multi-density crime hotspots and regressive models to automatically detect high-risk crime areas in urban environments, and to reliably forecast crime trends in each area. The algorithm result is a spatio-temporal crime forecasting model, composed of a set of multi-density crime hotspots, their densities and a set of associated crime predictors, each one representing a predictive model to forecast the number of crimes that are estimated to happen in its specific hotspot. The experimental evaluation of the proposed approach has been performed by analyzing a large area of Chicago, involving more than two million crime events (over a period of 19 years). This evaluation shows that the proposed approach, based on multi-density clustering and regressive models, achieves good accuracy in spatial and temporal crime forecasting over rolling prediction horizons. It also presents a comparative analysis between SARIMA and LSTM models, showing higher accuracy of the first method with respect to the second one.