Social capital variables at the neighbourhood level as predictors in a predictive policing model

Predictive policing models aim to predict crime events based on available crime and socio-economic data at a micro-geographic level. The aim of this study is to investigate the potential of including data on social capital, disorder and fear of crime for improving prediction performance of the predictive policing model. To this end, data is used from the Social capital and Well-being In Neighbourhoods in Ghent (SWING) survey, the Social Capital in Neighbourhoods (SCAN) project and the quality of life monitor from the city of Ghent. These datasets are based on extensive surveys at the neighbourhood level in Ghent and contain variables related to among others social capital, collective efficacy, social and physical disorders, and fear of crime. The prediction performance of two models are compared against each other: one model with the social capital variables included and one base model without these variables, with only basic crime and socio-economic variables included. The results of this analysis and its implications for the prediction performance of the predictive policing model will be discussed.