Estimation of Temporal-Spatial Forecasts for Crime Rates in Israel

• Activity Number: 674
• Type: Contributed
• Date/Time: Thursday, August 4, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Business and Economic Statistics Section
• Author(s): Yury Gubman*+ and Nir Fogel and Dmitri Romanov and Besora Regev and Shai Amram
• Companies: Israeli Central Bureau of Statistics and Israeli Central Bureau of Statistics and Israeli Central Bureau of Statistics and Israeli Police and Israeli Police
• Keywords: crime forecasting ; spatial models ; leading indicators ; time series

Abstract

Spatial allocation of limited police force requires extensive information and planning to optimize its effectiveness in terms of crime rates reduction. In most cases, crime activity is spatially and time dependent, and ignoring this fact may introduce a bias in estimators. Forecasts should internalize existing police data, economic, social and demographic variables. In this study, we analyze two major types of crime activities: violence and offences against property, for two geographical levels. Existence of spatial dependence in Israeli crime data has been tested. We propose temporal-spatial regression models for crime rates forecasting with different spatial metrics, and compare them with autoregressive OLS model which ignores spatial dependence. We use leading indicators method for medium and long term forecasting. For low geographical resolution, we apply indirect forecasting app