Parameter Estimation Of TV-INAR Model Driven By Autoregressive Coefficients

In our daily life,integer-valued time series data is common,such as the number of robberies in a certain area each year,and the number of patients hospitalized in a certain hospital each month.Therefore,the study of integer-valued time series is necessary and meaningful.Previously,scholars have conducted a lot of research and proposed a variety of models,including INAR(1)model,RCINAR(1)model and BAR(1)model,etc.According to realistic requirements,different models can be implemented on different types of data.The paper introduces the spatial TV-AR(time-varying autoregressive)model and the binomial sparse operator.After that,it cites Logistic transformation to study the basic properties,and analyzes parameter estimation of a class of autoregressive coefficient-driven TV-INAR(time-varying integer-valued autoregressive)models.Specifically,the thesis mainly uses conditional least squares estimation,conditional maximum likelihood estimation and Bayesian estimation to estimate the parameters in the model.As for the two parameters and ,it is found that the results of the three estimation methods are relatively accurate.By comparing the estimation effects,we recommend using the conditional maximum likelihood estimation method to estimate and the conditional least squares estimation method to estimate .As an application,we also carried out an example analysis of the proposed model to prove that the model has a certain significance on solving practical problems.