摘要
针对传统地理加权回归方法无法解决时空非平稳性的问题,该文提出了一种路网距离约束的时空地理加权回归方法。引入时间特性,进一步把握了不同因子在时空维度影响的分异性;以路网距离度量约束,提高模型解释力。以北京市城6区1980—2015年的1 632个住宅小区特征价格数据为例,通过与直线距离约束的常规地理加权回归方法等进行比较,采用各模型的AIC与拟合优度等指标对模型置信水平高低进行评价。实验结果表明,路网距离约束的地理加权回归模型不仅能够提高模型的拟合精度,还能更好地揭示房价在时间与空间方面的变化规律。
According to the fact that traditional geographical weighted regression methods cannot solve the problem of space-time non stationarity,this paper presented a geographically temporal weighted regression method constraint of road network distance.The introduction of time characteristics,further grasp the different factors in the spatial and temporal dimensions of the impact of the different;with the network distance constraints,the explanatory power of the model could be improved.Based on average housing price of 1 623 residential quarters within the main city of Beijing from 1980 to 2015,the confidence level of the model was evaluated using AIC and the goodness of fit index of each model and comparing with the normal geographically weighted regression method constraint of straight-line distance.The results showed that the geographically temporal weighted regression method constraint of road network distance can not only improve the fitting the accuracy of the model,but also reveal the change of housing price in time and space.
作者
王梦晗
刘纪平
王勇
罗安
徐胜华
WANG Menghan;LIU Jiping;WANG gong;LUO An;XU Shenghua(Shandong Agricultural University, Tai＇an, Shandong 271018, China;Chinese Academy of Surveying and Mapping, Beijing 100830, China)
出处
《测绘科学》
CSCD
北大核心
2018年第4期133-137,共5页
Science of Surveying and Mapping
关键词
GTWR模型
路网距离
时空非平稳性
北京房价
GTWR model
road network distance
space-time non stationarity
Beijing housing prices
