Computational Model for Spatio Temporal Crime Event Prediction

Abstract

With recent socio-economic developments the number of crime offences has risen drastically and the demographic factors are strongly cited as determinant of crime rate. In such scenario crime events usually related with a variety of crime socio-economic opportunity factors and hence the traditional system of criminal records has failed to maintain the required level of intelligence and predictive trends. In such irresolute environment the existing literature on crime prediction suggests to blend crime events with spatial-temporal analysis and use Generalized Linear Model for identification of the patterns within a criminal site selection criterion. Such model may improve quantitative prediction even during the temporal changes of criminals may transpires for certain sites. In this paper we enhance the Generalized Linear Model (GLM) for Crime Site Selection (CSS) and examine it for crime events using regression in association with big data technologies. This novel approach is based on standard deviation and autoregressive vector node has been proposed to find similar crime trends among various crime locations for criminal site selection and subsequently use this information for future crime trends prediction. This analysis presents a more significant insight into the scope and complexion of crime prediction with improved certainty. Keywords: crime prediction, spatio temporal, criminal site selection, generalized linear model, big data