Sex offenders and residential location: A predictive analytical framework

Despite the growing body of research dealing with sex offenders and the collateral consequences of legislation governing their post release movements, a complete understanding of the residential choices of registered sex offenders remains elusive. The purpose of this paper is to introduc… e a predictive analytical framework for determining which demographic and socioeconomic factors best forecast the residential choices of convicted sex offenders. Specifically, using a derived index of social disorganization (ISDOR) and a commercial geographic information system (GIS), we implement both linear statistical and non-linear data mining approaches to predict the presence of sex offenders in a community.  The results of this analysis are encouraging, with nearly 75% of registered offender locations predicted correctly. The implications of these approaches for public policy are discussed.