Comparative Study On The Spatial Variation Of Residential Land Price Based On GWR Model

The urban land price in dynamic change is important vane and leading indicator of land market.Land price shows the market operation,and to a certain extent it also embodies the development of urban social ecoaomic life.Different types of cities have different land price spatial variation characteristics.Knowing the spatial variation of urban land price and influences has important guidance function on improving land planning and utilization.In order to make an intensive study of the land price spatial variation of different cities,this paper makes a comparative study of the residential land price of Kunming,Guangzhou,Changchun City,these there cities have different properties,scales and geographical positions.Firstly,we collect residential land price data of these cities from 2012 to 2016,then take ESDA method to do Normal Distribution Test,Trend Analysis,Spatial Autocorrelation Analysis on the data,conclude the land price variation of each city.Secondly,we use Kriging Spatial Interpolation Technology to produce 2D and 3D graphs of land price to analyze the spatial variation rules.Last,we choose volume,building density,green space rate and area the four specific attributes of land as explanatory variables to build OLS model and GWR model,the tow models are used to analyze the relationship between the land rice and the four influence factors to explain the spatial variation.The main conclusion of this paper are as follows:(1)The residential land price data is not normal,but the data logarithm transform is;there are U and inverse U distribution on east-west and north-south derection,southeast-northwest and southwest-northeast direction;global spatial positive autocorrelation and local spatial agglomeraton exist between land price data samples of each city;all of the land price distribution trends of each city are similar to concentric circles,the land price declines from the city center to the edge region.(2)Spatial heterogeneity of residential land price in Kunming and Guangzhou City is caused by structural factors,but that of Changchun City is caused by both structural factors and random factors;the spatial autocorrelation is high in Kunming,middle in Guangzhou,weak in Changchun;all of the three cities'anisotropy are geometrical.(3)To all the cities,the result of Kriging Spatial Interpolation Technology shows the high land price area distributes in the center city district,radiate outside.(4)All of the cities' residential land price model test results show that GWR model has better fitting effect than OLS model.Comparing the result of each GWR model,we find that volume rate has the strongest influence to all cities' land price.Except Kunming,the factors in the descending order of influence degree are volume ratio,area,green space rate and building density.