Hypothesis Test Of Zero-one Inflated Poisson Regression Model

When we deal with some practical problems in the count data,there may be having excess zeros and excess ones.At this time,if we use the zero-inflated model to do the fitting,we will find that the fitting effect is not ideal.So we need to establish a zero-one inflated model.This paper focuses on the study of zero-one inflated Pois-son model,which deal with excess zeros and excess ones,and add to a covariate to obtain zero-one inflated Poisson regression model.Then a detailed derivation of the EM algorithm for parameter estimation of the zero-one inflated Poisson regression model.For the zero-one inflated Poisson regression model,the most important thing is to test whether there is excess zeros and excess ones in the given data.However,there are few test methods for this kind of problems.In this paper,based on the test ideas proposed by He et al.for the zero-inflated Poisson regression model,the zero-one inflated Poisson regression model test methods are proposed,that is Vuong test,Naive test,and He et al.test method.Then we conduct simulation studies to the three test methods.Through the type I error rate and test power,we can see that He test method can better than others in controlling the type I error,with stronger power.Finally,we apply He test for empirical analysis.