Spatio-Temporal Crime Prediction Model in Kathmandu valley using GIS

Abstract

Crime is an illegal act deviating from normal violation of the norms giving losses and harms for people. Social, psychological, economical and environmental factors are to be considered in crime issue. All these concepts affect occurrence of crime in different ways. Peoples who have role in crime prediction are police, local governments, law enforcement agencies and people exposed to crime and offenders. The spatio and temporal model is generated by using crime data for the year 2070 in Kathmandu police Headquarters. Methodology starts with obtaining clusters with K-mean, Nearest and Neighborhood(Nnh) and Spatial and Temporal Analysis clustering(STAC) algorithms. Above discussed clustering methods are compared in terms of number of crimes andland-use to select the most appropriate clustering algorithm. Crime data is divided into daily epoch, to observe spatio and temporal distribution of crime over the Kathmandu valley.To predict crime in time dimension a time series model (ARIMA) is fitted for each week day. Thespatial and temporal model of this thesis can give crime prediction in both space and time. Keywords:Spatial, temporal, ellipse, clustering, Spatial and Temporal Analysis of Crime (STAC),Euclidean distance, Geographical Information System (GIS), Autoregressive and Integrated Moving Average (ARIMA).