时空加权回归模型局部线性估计的渐近性质

Local Linear Estimation and Its Asymptotic Properties in Geographically and Temporally Weighted Regression Model

在模型分析中,对未知参数的估计及其相关性质的研究甚为重要。首先通过假定时空加权回归模型的回归系数是时空位置的函数,给出了时空加权回归模型的一种局部线性估计方法。其次,在假设I^V及引理1下,研究了由此估计方法得到的第p个系数函数估计量所具有的渐近条件性质。最后,证明了第p个系数函数估计量的渐近条件偏差和渐近条件方差。结果表明该估计具有良好的表现。

In model analysis,estimation of unknown parameters and its properties are very important.Firstly,this article assumes that the regression coefficients of geographically and temporally weighted regression（GTWR）model are functions of time and space location,then local linear estimation of regression parameters is proposed.Secondly,study the asymptotic properties of the p-th coefficient function estimator obtained according to this estimate under assumption I-V and lemma 1.Finally,the asymptotic bias and variance of the estimators of the proposed estimators are proved.The results show that the estimate has a good performance.