Penerapan Generalized Poisson Regression I Untuk Mengatasi Overdispersi Pada Regresi Poisson

(Studi Kasus: Pemodelan Jumlah Kasus Kanker Serviks di Provinsi Kalimantan Timur)

Abstract

Poisson Regression model is commonly used to analyze count data is assumed to have Poisson distribution where the mean and variance values are equal or also called equdispersion. In fact, this assumption is often violated, because the value of variance is greater than the mean value, this condition is called overdispersion. Poisson regression which is applied to the data that contains overdispersion will imply the value of standard error becomes underestimates, so the conclusion is not valid. One of the models that can be used for overdispersion data is Generalized Poisson Regression I (GPR I). This research discuss the handling of overdispersion on Poisson regression using GPR I, with case study modeling the number of cervical cancer cases in East Kalimantan in 2013. In this research GPR I models meet the criteria for suitability of regression compared Poisson regression models because it has a smaller AIC value.