Using geographically weighted regression for housing market segmentation

摘要: Urban real estate property values are mainly conditioned by several aspects, which can be summarised in two main classes: intrinsic and extrinsic ones. Intrinsic characters specific goods while features related to a diversity of goods. Therefore, there is an extremely close correlation between 'rigidity location' (fixed location) its value. Possibilities offered recent developments statistical techniques, principally geographically weighted regression (GWR), analysing housing market have given new impetus mass appraisal urban property. More particularly, has been adopted market, order identify homogeneous areas define the marginal contribution that single location (outlined these areas) gives value The model built on sample 280 data, trades residential units occurred 2008 2010 city Potenza (Basilicata, southern Italy). results territory zoning into areas, addition undoubted usefulness field valuations, useful implications terms taxation, programming territorial transformations checking ongoing or ex post planning decisions.