Research On Geographic Weighting Model Of Spatial Regression Stratification And Autocorrelation

Geographic Weighted Regression(GWR)is a modeling technique that effectively deals with regression analysis.It assumes that the regression coefficient is a function of the geographic location of the observation point.It allows different geographical locations to have different spatial relationships and solves the object of study.Spatial non-stationarity problem.The spatial econometric theory studies spatial relations of spatial objects and the heterogeneity of structures in a crosssectional data regression model by using spatial adjacency matrices.This dissertation combines spatial measurement of GWR model and two weight matrices.The economic model was obtained by spatially stratified autocorrelation geographically weighted regression model(SSAGWR),which solves the spatial non-stationarity problem of the research object well,and at the same time,the hierarchical autocorrelation model can be applied to the individual level.Simultaneous analysis of data at the level of ethnic groups and the effects of individual variables and variables at the group level are tested simultaneously in the model.Therefore,from the perspective of statistical analysis techniques,the problems encountered by traditional analysis methods in analyzing multi-layered data can be Solved by SSAGWR model.This dissertation first introduced the GWR model and the spatial econometric model.After that,it introduced the superiority of the two-stage estimation of the SSAGWR model in dealing with multi-layer data structures.Then it analyzed the parameter estimation and statistical inference of the SSAGWR model.The simulation test compares the accuracy of the GWR model,draws a comparative surface map,and compares the mean of the deviation squared with the corresponding results of the GWR model to illustrate the macroscopic and microscopic aspects of the SSAGWR model in the study of different levels of spatial data.Spatial dependence.In recent years,due to the diversification of data acquisition methods and the increasing emphasis on “spatial roles”,“positions and different levels” in many disciplines,the spatial econometric model has received in-depth research and universal attention.The two-stage estimation method of the coefficient model combined with the spatial econometric model compares the difference between the SSAGWR model and the classical GWR model from a theoretical perspective,highlighting the theoretical innovations of this paper.Innovation is mainly reflected in: The classic GWR In the model,when spatial correlation is analyzed,it only contains one level of spatial correlation.Based on a detailed explanation of geographically weighted regression model and spatially stratified econometric model,this paper theoretically studies two levels of spatial dependence and Heterogeneity issues.