Spatial heterogeneity of influential factors of urban housing prices based on multiscale geographically weighted regression models: A case study of the main urban area of Guilin city(基于多尺度地理加权回归模型的城市住宅价格影响因素空间异质性)

Abstract

Based on second-hand housing prices data in Guilin city in October 2022, this paper investigates the spatial heterogeneity of factors influencing housing prices at different scales using the multi-scale geographically weighted regression (MGWR) model. The results indicate that: (1) Significant spatial autocorrelation is observed in housing prices in Guilin City, and the local spatial autocorrelation reveals a distinct clustering pattern characterized by predominant "high-high" and "low-low" clusters, with smaller occurrences of "high-low" and "low-high" clusters. (2) Compared to ordinary least squares (OLS) model and geographically weighted regression (GWR) model, MGWR model demonstrates higher coefficients of determination, enabling more accurate measurement of the scale and spatial variability of variable effects on housing prices. (3) The age of housing, education/medical and business facilities serve as global variables; greening rate, public transportation infrastructure, railway facilities, living amenities, and tourist attractions exhibit moderate levels of spatial heterogeneity; external transportation facilities display greater spatial heterogeneity, exerting significant and distinctly varied effects on housing prices within local areas. (4) The age of housing, living amenities, and tourist attractions negatively influence housing prices negatively; education/medical, and business facilities positively influence housing prices; while other variables exhibit varying inhibitory or appreciative effects on housing prices across different spatial locations.(基于桂林市2022年10月二手房价格数据，结合多尺度地理加权回归（multi-scale geographically weighted regression，MGWR）模型，分析不同尺度下城市住宅价格影响因素的空间异质性。结果表明：（1）桂林市住宅价格呈显著空间自相关，局部空间自相关结果表明，桂林市住宅价格空间呈显著的集聚分布，以“高-高”和“低-低”集聚为主，“高-低”和“低-高”集聚范围较小；（2）相较于普通最小二乘法（ordinary least squares，OLS）和地理加权回归（geographically weighted regression，GWR）模型，MGWR模型的R2更高，能更准确地测度变量对住宅价格的影响程度和空间差异性；（3）住宅房龄、教育医疗和商务设施为全局变量，绿化率、公共交通设施、铁路设施配套、生活配套和旅游景点的空间异质性处于中等水平，对外交通设施的空间异质性较大，在局部范围内对住宅价格有显著影响且差异明显；（4）住宅房龄、生活配套和旅游景点对住宅价格具有负向影响，教育医疗和商务配套对住宅价格具有正向影响，其余变量在不同空间区位上对住宅价格表现出抑制或增值作用。)

Keywords

• housing prices(住宅价格)
• influencing factors(影响因素)
• spatial heterogeneity(空间异质性)
• multi-scale geographically weighted regression(多尺度地理加权回归)
• guilin city(桂林市)