Error Time Series Of Semi-parametric Regression Model Of The Knife Wavelet Estimation

In this paper, research in the fixed design semiparametric regression model:The semi-parametric regression model focused on the parameters of the mainpart of the component parts of the information, have a good explanatory power.Wavelet analysis is the mathematics of a rapid development of a new direction, a newmathematical method is developed by the Fourier analysis, and also has a profoundtheory and application of a wide range of double meaning. Since the wavelet basis is awavelet function through the pan and scalable, and therefore has a simple, flexible,random features, but also has multi-resolution analysis function. A new and moresuperior and convenient analytical tool for many applications. In this article, weconsider the random error for the case of a smooth sequence, which containsAR (p)， MA (q)， ARMA (p,q)a variety of time series models. The second chapter,the premise of the theorem made about Introduction, Chapter III, on the basis of theparameters of the model, wavelet as the integral kernel to construct the weightfunction of the unknown parameters for least squares estimation and knife typeestimated in this chapter the following conclusions:(1) proved that the semi-parametric regression model knife wavelet estimationasymptotic normality.(2) discuss the asymptotic variance of the estimated function knifem β∧n(3) discuss the nature of the limits of the incremental variance.