北京市二手住宅价格影响机制——基于多尺度地理加权回归模型(MGWR)的研究

文章基于多尺度地理加权回归研究北京市2011—2017年二手住宅交易的价格特征,结果表明:①以往基于经典地理加权回归模型的研究可能存在一定的不稳健,而多尺度地理加权回归可以将不同变量对于因变量的影响尺度反映出来,其回归的结果更为可靠。②北京房价对区位因素非常敏感,且存在高度的空间异质性,区位的影响尺度是所有变量中最小的,接近于街道尺度。而卧室数量和到最近地铁站的距离为全局尺度的变量,在空间上的影响较为平稳。到公交站的距离、到小学的距离、建筑结构和装修状况对于房价的影响不显著。其他显著的变量均存在一定的空间异质性,其空间尺度由小到大分别为成交时间、面积、楼龄、楼层、朝向。③区位、朝向、卧室数量、成交时间均正向影响房价,而面积、楼龄、楼层、到地铁站的距离负向影响房价。所有影响因素中区位是影响房价的最主要因素,其次是成交时间朝向。面积成交时间、朝向和到最近地铁站的距离影响较大,所在楼层、卧室数量对于房价的影响较小,而面积和楼龄的影响最弱。

On Hedonic Price of Second-Hand Houses in Beijing Based on Multi-Scale Geographically Weighted Regression:Scale Law of Spatial Heterogeneity

A large number of empirical researches show that there is obvious spatial heterogeneity in the housing price influence mechanism.Although classic geographically weighted regression(GWR)can solve part of the spatial heterogeneity problem which cannot be handled by traditional linear regression models,it ignores the scale problem of spatial heterogeneity of different influencing factors and causes large estimation errors.Multi-scale geographically weighted regression(MGWR)improves classical GWR by allowing the bandwidths of each variable to be different,thereby obtaining more credible estimation results and giving the scale of influence of different variables.Based on MGWR,this paper studies the price characteristics of second-hand residential transactions in Beijing from 2011 to 2017.The results show that:1)Previous researches based on classic GWR may not be robust.MGWR can separate different influence scales of the independent variables,and MGWR result is more reliable.2)Beijing second hand house prices are very sensitive to location factors,and there is a high degree of spatial heterogeneity.The scale of location impact is the smallest of all variables and is close to the street scale.The number of bedrooms and the distance to the nearest subway are variables on a global scale,and the influence on space is relatively stable.The distance to the bus station,the distance to the elementary school,the structure of the building and the condition of the decoration have no significant effect on house prices.Other significant variables have certain spatial heterogeneity,and their spatial scales from small to large are transaction time,area,building age,floor,and orientation.3)Location,orientation,number of bedrooms,and transaction time all positively affect house prices,while area,age,floors and distance to the subway station negatively affect house prices.In all the influencing factors,location is the most important factor affecting house prices,followed by the direction of transaction time.Area,transaction time,direction,and distance to the nearest subway have key impact.Floor and the number of bedrooms have a smaller impact on house prices,while the area and building age have the weakest impact.