时空加权回归模型的非平稳性检验

The Nonstationarity Tests of Geographically and Temporally Weighted Regression Model

时空数据具有空间非平稳性和时间相关性的特征,单纯考虑空间因素或时间因素,用地理加权回归模型或时间加权回归模型来拟合时空数据,其分析结果不能全面反映时空数据的真实特征.时空加权回归模型通过在线性回归模型中假定回归系数为地理位置和观测时刻的函数,将数据的时空特性纳入到模型中,为探索回归关系的时空平稳性创造了条件.基于加权最小二乘估计理论,给出了时空加权回归模型回归关系的空间平稳性检验和时间相关性检验方法.

There are the characteristic features of spatial nonstationarity and temporal correlation for the spatio-temporal data. If we consider only the spatial factor or temporal factor to fit the spatio-temporal data by geographically weighted regression model or temporal weighted regression model, the analysed results can not reflect the true characteristics of spatio-temporal data. Geographically and Temporally Weighted Regression Model assumes that the regression coefficients are the functions of geographical position and observation time in linear regression model, and the spatio-temporal characteristics of data are involved in the model,thus creating conditions for exploring the spatio-temporal nonstationarity of the regression rela- tion. The methods of spatial nonstationarity test and temporal correlation test of the regression relation of geographically and temporally weighted regression model are introduced here on the basis of the theory of weighted least squares estimate.