Study On Spatial Distribution Change And Trend Of Urban Housing Price Based On Multi-scale GTWR

The housing problem is the focus of current social concern.Housing price is not only affected by macro economy,market supply and demand,national policies and other aspects,but also affected by micro aspects such as location and housing characteristics.Therefore,the study of housing transaction price is of great significance to urban planning and housing security.From the perspective of geographical location,this attempts to explore the spatial-temporal distribution characteristics and variation rules of housing prices in typical mountain cities by using Kriging interpolation and other spatial analysis methods,and uses the geographical weighted regression method and its expansion model to study the price characteristics of housing prices in mountain cities.The research area selected in this paper is the main urban area of Chongqing,a typical mountain city.The research of this starts from collecting and collating the transaction data and basic geographic data of second-hand traded housing in the main urban area from 2018 to 2021,and establishing a complete geospatial database.Then,through GIS exploratory spatial data analysis,the spatial and temporal distribution characteristics of urban housing prices are summarized.At the same time,Kriging interpolation method is used to find out the main center and sub-center of housing price from 2018 to 2020,and analyze the reasons for their changes.Then,the purchase intention survey for citizens,preliminary screening of housing price factors.The conventional linear regression(OLS)model is built,and the coefficient of the influence factors is obtained exploratively.Six independent variables of housing price are obtained by removing the influence factors with smaller coefficients.Finally,the geographical weighted regression(GWR)model,multi-scale geographical weighted regression(MGWR)model and multi-scale spatiotemporal geographical weighted regression(MGTWR)model are constructed respectively.The fitting accuracy of the models is compared and the regression results are visualized,and the scale dependence of the regression results is analyzed.The main findings are as follows:1.From 2018 to 2020,the main price center of Chongqing’s main urban area has been "Jiangbeizui",but the sub-center of price has changed: In 2018,the price subcenter is mainly concentrated in the "Daping" and "Zhaomushan Park" area.In 2019,the price subcenter changes from "Daping" to the "Central Park" area of the airport Group.In 2020,the price subcenter remains the same as that in 2019,and the two areas "Xiyong Group" and "Photoelectric Park" are added to the price subcenter.2.After preliminary screening of the impact factors and OLS model analysis,six variables are finally selected as the impact factors of the housing price in Chongqing.The GWR,MGWR and MGTWR models are used to fit the variables,and the fitting accuracy is 52.21%,68.80% and 82.60%,respectively,which confirmed that the MGTWR model greatly improved the fitting ability of housing prices in the main urban area of Chongqing.3.According to the regression coefficient of MGTWR model,the effect of urban location on housing price is the most prominent.In 2018,the degree of old and new housing has the largest impact on housing price among independent variables.In 2021,the surrounding environment of the community has the greatest impact on housing price among independent variables.In addition,it is found in the regression results of the model that the housing price has a certain scale dependence: in the spatial scale,the 6independent variables all have large spatial heterogeneity,among which the degree of medical facilities and decoration has the strongest spatial heterogeneity.From the perspective of time scale,the time heterogeneity of decoration degree and medical facilities is large,indicating that the same variable plays a different role in the same location over time.The time scale of the remaining variables is close to 4,and the time heterogeneity is small,indicating that the effect difference of variables is small,which has certain reference significance for housing price forecasting and urban management.