ZERO-INFLATED POISSON REGRESSION ANALYSIS ON CLAIM FREQUENCY OF MOTOR VEHICLE INSURANCE

• Main Author: SUSILO, HENDRAWAN
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/65454
• Institution: Institut Teknologi Bandung

Poisson regression is the most frequently used model for claim frequency. The basic assumption of the Poisson distribution is that the variance and the mean have the same value (equidispersion). However, in the insurance industry, count data often experiences overdispersion due to an excessive number of zeros. For such cases, the Zero-inflated Poisson regression model is more suitable to use. In this study, AIC, BIC, and Vuong test were used to compare the two models. The Zero-inflated Poisson regression model was applied to a case study of the claim frequency of motor vehicle insurance. The frequency of insurance claims observed in this study is known to experience overdispersion as many observations are zero data, meaning that a lot of policyholders do not file a claim. The results of this study indicate that the Zero-inflated Poisson regression model is more suitable for use in data on claim frequency of motor vehicle insurance compared to the Poisson regression. Furthermore, this study also examines the ability of Zeroinflated Poisson (ZIP) regression to see the limit of zero probability in overcoming overdispersion in the data on claim frequency of motor vehicle insurance. Based on the simulated data, it is found that the ZIP regression stops overcoming overdispersion under conditions with a probability of ???? = 0.7.