都市新区住宅地价空间异质性驱动因素研究 ——基于空间扩展模型和GWR模型的对比

以南京市江宁区为例,基于2004~2011年住宅用地出让数据,利用空间扩展模型和GWR模型对都市新区住宅地价空间异质性及其驱动因素进行研究。结果表明：① 空间扩展模型与GWR模型分别可解释采样区63%、61%的住宅地价变化,较全局回归模型(47%)有显著提升,更有利于研究土地市场的空间异质性。② 空间扩展模型可有效表征各解释变量及其交互项对住宅地价作用的空间结构总体趋势,其拟合效果相对较优。GWR模型则在局部参数估计方面存在优势,借助GIS可将各变量的地价作用模式可视化,从而比空间扩展模型更能有效刻画住宅地价影响因素的空间非平稳性特征,各因素对地价的平均边际贡献排序为水域> 地铁> 大学园区> CBD> 商业网点> 医院,且商业网点、 医院系数值具有方向差异性。③ 距地铁站点、水域、大学园区以及CBD的距离是研究区住宅地价的关键驱动因素,各自存在特有的地价空间作用模式,可为研究区住宅土地市场细分提供科学依据。

Drive Pattern on the Spatial Heterogeneity of Residential Land Price in Urban District: A Comparison of Spatial Expansion Method and GWR Model

Based on the spatial properties of the database and the final retained Jiangning District Residential land transfer data from 2004 to 2011, spatial expansion method and geographically weighted regression (GWR) model are applied to simulate the spatial heterogeneity of residential land market in urban district. The influencing factors of residential land price were also tested and analyzed. The results show the following aspects.1)Spatial expansion method and geographically weighted regression (GWR) model can be well applied to simulate spatial heterogeneity of land market in target area. The model could respectively explain 63% of the price changes of residential land and 61% of the price changes of residential land. The interpreting abilities improve significantly than that based on global regression model (47%). Both explanations capacity increased by 16%, 14% and spatial expansion method is slightly better than geographically weighted regression (GWR) model. 2) Spatial expansion method can effectively characterize the spatial structure of the overall trend, which is reflected from the explanatory variables and their interaction term effects on residential land. Geographically weighted regression (GWR) model has advantages in terms of the local parameter estimation. It can make the mode of action of each variable premium visualization by means of GIS. This is a strong rebuttal of the traditional assumptions that hedonic price model has coefficient stability. Overall, the spatial expansion model fits relatively better results. Compared with spatial expansion method, geographically weighted regression (GWR) model can more effectively depict spatial non-stationarity of the influencing factors. In the geographically weighted regression(GWR) model, the order of the average marginal contribution on the land premium from high to low is the distance from water, subway, college and CBD, facility, hospital. Additionally, two variables, the distance from facility and hospital, have the directional difference. 3) The distance from subway, water, university and CBD all have positive effect on marginal residential land price in the entire sample area. They are the key driving factors of residential land price .Each of the affecting patterns has a unique land premium space mode of action. Therefore it can provide scientific basis for segmentation of residential land market in target area. Marginal price effect of waters in residential areas of rapid urbanization is generally greater than industrial areas surrounding. The construction of the subway greatly contributed to construction land expansion of Crisscross and upgrading of residential land price. The construction of University City is also an important strategy for urban development in Urban District, The higher the density of University City, the more significant Its role in promoting residential land price. The marginal price effect of entral business dstrict(CBD) on residential land is progressively decreasing trend from the periphery inward City, but it still has upgrading effect on the surrounding residential land price.