Zulkifli, Malina and Ismail, Noriszura and Razali, Ahmad Mahir (2012) Zero-inflated regression models with an application to vehicle theft count data. In: Technology Science, Social Sciences, Humanities International Conference 2012, 14th to 15th November 2012, Langkawi, Kedah, Malaysia. (Unpublished)

PDF Restricted to Registered users only Download (346kB) | Request a copy

Abstract

Poisson regression model has been widely used for modeling claim count data in actuarial and insurance literatures.However, in several cases, claim count data often have excessive number of zeros than are expected in the Poisson model. In that case the Poisson regression may underestimate the standard errors and giving misleading inference about the regression parameters. This paper aims to apply the zero-inflated regression models on vehicle theft crime data.These zero-inflation phenomenon is a very specific type of over dispersion and zero-inflated Poisson (ZIP) regression model has been suggested for handling zero-inflated data.If the crime count data continue to suggest additional over dispersion, the alternative models the zero-inflated negative binomial-1 (ZINB-1) and the zero-inflated negative binomial-2 (ZINB-2) will be fitted on the private car theft claim count data.In addition, two different forms of link function will be used in the fitting procedure of the zero-inflated regression models, producing different estimates for each model.The results of this study indicate that the ZINB-2 models is better compared to ZIP regression model for handling zero-inflated and additional overdispersed crime count data.

Item Type:	Conference or Workshop Item (Paper)
Additional Information:	Organized by: Universiti Teknologi Mara Kedah
Uncontrolled Keywords:	vehicle theft, count data, crime, zero-inflated poisson, zero-inflated negative binomial
Subjects:	Q Science > QA Mathematics > QA76 Computer software
Divisions:	College of Arts and Sciences
Depositing User:	Mrs. Malina Zulkifli
Date Deposited:	06 Jul 2015 06:44
Last Modified:	06 Jul 2015 06:44
URI:	https://repo.uum.edu.my/id/eprint/14753

Actions (login required)

View Item