具有一阶自回归误差的半参数可加模型的估计

Estimation of the Semiparametric Additive Model with First Order Autoregressive Errors

研究具有一阶自回归误差的半参数可加时间序列模型的估计问题.假设回归函数来源于某个参数分布族,采用局部二次拟合准则,结合非参数核函数方法,给出参数向量的估计量和回归函数的半参数估计量.在一定正则条件下,证明估计量具有相合性.通过模拟研究与实证分析检验方法的有效性和可行性.

Estimation of the semiparametric additive time series model with first order autoregressive errors is mainly studied.We suppose that the regression function has a parametric framework,and through the local L_(2)-fitting criterion,parametric vector estimators and semiparametric estimators of the regression function can be given under the use of nonparametric kernel function method.Furthermore,under certain regular conditions,the consistency of the estimators is proved.Finally,simulation research and empirical analysis are presented to evaluate the effectiveness and feasibility of the proposed method.