NEGATIVE BINOMIAL REGRESSION AND GENERALIZED POISSON REGRESSION MODELS ON THE NUMBER OF TRAFFIC ACCIDENTS IN CENTRAL JAVA

Abstract

Traffic accidents that always increase along with the increasing population growth and the number of vehicles impact the national economy. The number of traffic accidents is a count data that a Poisson distribution can approximate. The Poisson regression model often found violations of the overdispersion assumption by modeling the factors that affect the number of traffic accidents. Alternative models proposed to overcome the emergence of overdispersion in the Poisson regression model are the Generalized Poisson Regression and Negative Binomial Regression Models. Based on the analysis results, it was found that the overdispersion assumption violates the Poisson regression model, and the Generalized Poisson regression model is the best because it has the smallest AIC value of 485.50. Factors that significantly affect the number of traffic accidents in Central Java Province are the percentage of adolescents and the percentage of accidents occurring in the road area of the district/city.