Geographically weighted regression (GWR) is a simple but powerful method for exploring spatial relationship non-stationary. In GWR, all coefficients vary over space, and the parameter estimates are made using an approach in which the contribution of a sample to the analysis is weighted based on its spatial proximity to the specific location under consideration. Data from observations close to the location under consideration are weighted more than data from observations far away. And parameter estimates are local rather than global parameters at each point on study region. But in some situations not every coefficient in a model varies geographically, so mixed geographically weighted regression, in which some coefficients are termed global while the remaining coefficients are termed local, should be considered. Undoubtedly, ordinary algorithm of parameter estimation of GWR is not directly used in mixed GWR. In this paper, based on some work with GWR technique, an iterative algorithm is developed to estimate global coefficients and local coefficients in GWR, and this method is further tested by using average sold prices of house blocks in Shanghai, China. The experiment proves that mixed GWR is more appropriate and stable for the local coefficients estimates than GWR, although it requires a greater computational effort.