Research On Residential Land Price Evaluation Model Based On BP Neural Network

With the rapid development of social economy and the improvement of infrastructure construction,the quality of urban land and the level of land price are also changing.Urban land price evaluation is becoming more and more important to the government in planning the land market,land grading and optimizing the use of land resources.Only by observing the urban land price in real time and accurately can we timely reflect the spatial distribution law and actual land price level of land price and meet the needs of the development of land market.The traditional evaluation methods have great restrictions on the use of urban land price evaluation,and the evaluation validity and accuracy need to be improved.Therefore,this thesis attempts to introduce artificial intelligence technology into the field of land price evaluation for method exploration,in order to promote the development of land evaluation in our country.Based on BP neural network method and Arc GIS technology,this thesis constructs the spatial database of urban residential grid land price influencing factors.Based on the advantages of nonlinear mapping ability and generalization ability of BP neural network method,this thesis constructs the land price evaluation model of the relationship between urban residential grid land price and its influencing factors,and makes an empirical analysis with the main urban area of Xuzhou as the research area,The applicability and advanced nature of BP neural network for city grid grid land price evaluation is verified.Results:(1)according to the classification system of influencing factors of urban residential land price and relevant theories of land price,the four influencing factors of Commerce,transportation,infrastructure and environment are scientifically selected and refined into 10 influencing factors such as commercial service prosperity.Then,based on the cost distance mapping and other tools provided by Arc GIS platform,the spatial data system of influencing factors of land price in Xuzhou is successfully established.(2)L-M algorithm is introduced to optimize and improve the limitations of standard BP network.The improved BP neural network is used to scientifically simulate the nonlinear relationship between residential grid land price and its influencing factors.The scientificity and feasibility of modeling are expounded in theory.(3)Taking the main urban area of Xuzhou as an example,the BP neural network land price evaluation model is constructed by using 150 groups of sample data.The average relative error of the predicted land price of the test set is 2.528%.Compared with the evaluation results of the regression model and the benchmark land price correction method,the applicability and accuracy of BP neural network in the field of land price evaluation are verified.There are 14 figures,21 tables and 82 references in this thesis.