Nonlinear Time Series Model With Exogenous Variables And Its Application

The nonparametric method and semi-parametric method were focused in nonlinear time series analysis. The theories of nonparametric or semiparametric regression have been developed ,but there is few application in practical problems.So in this paper,one tried to apply those theories or methods of nonparametric and semi-parametric regression to nonlinear time series model with exogenous variables.So the additive coefficient model for the data of our country civil income and consumption,the semi-parametric autoregressive model and function coefficient autoregressive model for the data of our country population and GDP and the function coefficient model for the data of SARS daily increase and daily temperatureare instructed .For concrete as follows:1. For the data of our country civil income and consumption, this paper instructed the additive coefficient model by polynomial spline estimation method.the result is that this model is better than the linear autoregressive model and the nonparametric model.2. By kernel method and polynomial spline estimation, the semi-parametric autoregressive model and the function coefficient autoregressive model separately are constructed.3. One constructed function coefficient autoregressive model about the data of SARS daily increase quantity and daily temperature with two step local linear method.