An Early Warning System for the Prediction of Criminal Careers

Dismantling networks of career criminals is one of the focus points of modern police forces. A key factor within this area of law enforcement is the accumulation of delinquents at the bottom of the criminal hierarchy. A deployed early warning system could benefit the cause by supplying an automated alarm after every apprehension, sounding when this perpetrator is likely to become a career criminal. Such a system can easily be built upon existing, strategic, analysis already performed at headquarters. We propose a tool that superimposes a two-dimensional extrapolation on a static visualization, that describes the movement in time of an offender through the criminal spectrum. Using this extrapolation, possible future attributes are calculated and the criminal is classified accordingly. If the predicted class falls within the danger category, the system could notify police officials. We outline the implementation of such a tool and highlight test results on a criminal record database. We also touch upon the issue of estimating the reliability of a single, individual criminal career prediction.