Research On The Application Of Spatial Measurement Model In Real Estate Batch Appraisal

In recent years,the real estate appraisal industry is undergoing tremendous changes and innovations.Big data has become more and more closely related to people’s lives.The demand for real estate appraisal is also increasing,and the relevant theories and practices of real estate batch appraisal have been developed under this condition.The issue of housing prices has always been a topic of great concern to the government,society and academia.In the real estate batch appraisal,the core idea is based on the real estate in a certain area using scientific and technological means to efficiently,accurately and fairly complete its appraisal.However,the new technologies and methods of real estate appraisal currently used in practice,case studies on different real estate markets and different sectors in the same market,and large data batch accurate appraisal methods are lacking.This paper takes Panlong District of Kunming City as an example to study the applicability of the batch evaluation model and analyze the main influencing factors of residential real estate.In terms of empirical research: First,quantify the characteristics of residential samples based on the selected feature variables through Baidu Map’s GIS technology,establish a second-hand house sample database in Panlong District after excluding outliers and missing values,and perform descriptive statistics on the data.Analysis;Secondly,based on 340 sample data,three common feature price models were established and the regression effects of the three functional forms were compared.The comparison found that the model fitting effect under the full logarithmic function form was the best;Stepwise regression is performed on the mathematical model,and there are 10 characteristic variables in the optimal regression subset;again,spatial factors are introduced on the basis of the traditional characteristic price model(full logarithmic form)to carry out maximum likelihood estimation of housing prices to establish a spatial measurement model;Finally,40 samples of Panlong District,Kunming City,which did not participate in the model construction,were substituted into the traditional feature price model and the spatial error model for evaluation.The accuracy of the evaluation results under the two models was compared and the pros and cons of the model were judged.Through the empirical analysis of the sample data of second-hand housing in Panlong District,Kunming City,the main conclusions are as follows:(1)Second-hand housing prices in Panlong District,Kunming City are spatially dependent,and the results of AIC criteria and Log likelihood indicate spatial errors.The model has better explanatory power for the selected area samples.In the error analysis of the evaluation results,the error between the evaluation value calculated by the spatial error model and the true value is smaller and the ratio test is within a reasonable range;(2)For the real estate batch evaluation of second-hand housing sample data in Panlong District,Kunming City On the one hand,the spatial error model overcomes the characteristics of the traditional feature price model that does not consider the spatial autocorrelation between samples,and on the other hand can improve the efficiency and accuracy of real estate batch evaluation.Therefore,in the practice of real estate batch appraisal,the spatial effect between sample data should be properly considered.