The BIGDATPOL research programme includes building and evaluating a machine learning model that uses (big) data sources (e.g. crime data available in police databases, crime opportunity indicator data such as number of shops, street connectivity, socio-economic data such as age, median income) to anticipate the risk of where and when a residential burglary is likely to occur. BIGDATPOL is based on previous research, such as key contributions on theory testing of crime concentration at micro places, crime indicators, deterrence, and crime prevention interventions.

A suitable model is intended to support law enforcement and crime prevention via an informed allocation of police patrols and the suitability should be determined through robust evaluations. However, there are currently a lack of evaluation studies of big data policing applications, highlighting the need to address this gap.

The BIGDATPOL team at Ghent University will build, test, and evaluate the model based on three core dimensions – i.e. statistical & methodological (e.g. performance), criminological (e.g. user experience) & economic (e.g. cost-benefit analysis), and ethical (risk of bias and stigmatisation) & legal (e.g. data protection). Using a mixed methods approach, the holistic assessment of the model will thereby both facilitate a better understanding of the utility of big data policing models in crime prevention while simultaneously enhancing the model itself.