Traditionally, the relationship between housing amenity values and property prices has been investigated through the global model (OLS), in which spatial heterogeneity is largely ignored in the regression procedure. This study aimed to determine the relationship between housing prices and ‘green’ amenity values brought by open spaces in terms of presence, quantity and quality. The spatial variations in the spatial statistical model were explored through the Geographically Weighted Regression (GWR). GWR can effectively calibrate spatial variations and generate local regression models fitted at each location with greater accuracy and lower residual as compared with global regression results. The statistical results demonstrated that GWR significantly improved the Adjusted R Squared from 0.51 of the OLS model to 0.71, and it also decreased AICc by approximately 1.5%. Furthermore, the GWR model successfully increased the accuracy of the model and decreased the residual by over 53.8% as compared with the OLS. When used in conjunction with the geographical information system (GIS), the estimated coefficients and other important statistics of model variables could also be visualized for further analysis. The analysis indicated that the quantity of open space is more important than the quality of open space. With the existence of open space, the mean housing prices in specific areas of the City of Edinburgh could be influenced by approximately 6.5%. In contrast, no explicit relationship between housing prices and the quality of open space was observed.