Comparison of INAR(1)Models Based on Negative Binomial Thinning Operators and Extended Negative Binomial Thinning Operators
This article uses the method of conditional maximum likelihood estimation to compare the parameter estimation of the INAR(1)model based on the generalized negative binomial thinning operator and the INAR(1)model using the preset innovation distribution method.By conducting numerical simulations on two models with different distribution of innovation terms and comparing the numerical simulation results of the two models,it is intuitively shown that the INAR(1)model based on negative binomial thinning operator does not have a probability mass at point x=0,and therefore all edge distributions of the model do not exist.This proves that the INAR(1)model based on negative binomial thinning operator using the preset innovation distribution method does not exist,This also indicates the necessity of the generalized negative binomial thinning operator in constructing the INAR(1)model using the preset innovation distribution method.
negative binomial thinning operatorbinomial thinning operatorINAR(1)Modelcondition maximum likelihood estimation
万方数据
维普
