考虑自回归模型Yt=θ^TXt+g(Zt)+εt,t=1,…,n,其中Xt=(Yt-1,…,Yt-d)^T,Zt为实值外生随机变量,θ=(θ1,…,θd)^T为待估参数向量,g为未知非参数光滑函数.基于多项式样条方法,在一定的条件下,给出了θ的估计的渐近正态性,得到了g的估计的收敛速度.模拟例子验证了所得的理论结果.
Consider the autoregressive modelYt=θ^TXt+g(Zt)+εt,t=1,…,n,whereXt=(Yt-1,…,Yt-d)^T,Ztis a real value exogenous random variable, θ = (θ1,... , θd)^T is an unknown parameter vector to be estimated and g is an unknown nonparametric smooth function. Based on polynomial spline estimation, asymptotic normality of estimator of θ is given and convergence rate of estimator of g is obtained under some mild conditions. Main results are illustrated by a simulation examDle.