Modelling the influence of location on value

Abstract

Describes an attempt to predict the value of private properties in the Stafford area from information typically available about the properties’ physical characteristics and location. Discusses the factors influencing property values, including the complicating effect of location, and attempts to build a predictive model using the statistical technique of multiple regression analysis, based on physical characteristics only and ignoring the effect of location. Makes further refinements to the predictions produced by incorporating an additional location effect modelled by means of surface‐fitting techniques within a geographical information system. The results, many of which are presented graphically, are shown to be a significant improvement on the location‐blind model. Offers suggested directions for further research and development of the technique.