Geographically weighted regression model is widely used in spatial data anal-ysis. Mixed geographically weighted regression model as an extension of the ge-ographical weighted regression model has very important actual value. In this paper, analysis for mixed geographically weighted regression is divided into two parts:In the first part, the precision and coefficients of parameter constraint estimation of Mixed geographically weighted regression prove to be well. Baced on F-approximation methods we built model test statistics and compare the constraints estimations solved by Two-step estimation method with the Back-fittings'. Test the accuracy and validity of the test statistics; ON the other hand, error with in mixed geographically weighted regression model were discussed, according to Moran's I, Geary's C statistics and F approximation method to construct the spatial autocorrelation of the test statistic. Finally, a numerical simulation of the test statistics carried out to verify. |