Crime Data Mining: Combining Socio-economic and Spatial Analysis

Public security is an important issue for a society. With the massive increase in electronic data availability over the last few years, characterising and predicting crime has become a task that can be approached using different data mining techniques. However, previous studies have concentrated on the spatial patterning of crimes without investigating potential predictors or causes. In this paper, we use a range of data mining techniques to analyse criminality using a data base of crimes from the municipality of General de Escobedo in Nuevo León, México. We show that different types of crime domestic violence, residential robberies and business robberies have quite different profiles, both from the point of view of the characteristics of the robberies themselves and from the underlying socio-demographic and socio-economic factors that influence them. We create predictive models for these three crime types and discuss how the results can be used to predict and reduce crime risk.