Analysis Research Of Apartment Price Based On Improved BP Neural Network

With the rapid development of the market economy, China's real estate market increasingly actives, gradually towards completion.Related to this real estate appraisal business plays an important role in real estate market, rapid develops and expands,service scope increasingly broads. The real estate appraisal plays an inestimable role in promoting real estate prices normalization, protecting fair trading and building up a healthy market system. Current research on the real estate appraisal is still in exploration stage, rarely establishs mathematical model to quantitative assessment, so it has important theoretical and practical significance to actively look for more scientific evaluation methods.This paper first systemly discusses the current real estate price evaluation used in broad method:the market comparison, cost method and income method of these valuation methods, systemly analysises the characteristics, application conditions and range of these methods. Analyzes the deficiencies of the traditional appraisal method in dealing with the nonlinear relationship between the real estate influencing factors and the transaction price.Secondly, introduces the basic ideas of artificial neural network, network structure and its training algorithm, discusses and derivates BP algorithm,summarizes and analyzes the improved algorithm of BP neural network. Artificial neural network with nonlinear function dynamic processing power, self-organizing, adaptive and self-learning function, can be used for modeling tools.Thirdly,based on the predecessor's research, demonstrates the feasibility of artificial neural network theory in real estate appraisal, uses the super processing ability of the nonlinear relationship to construct the neural network model of the real estate appraisal from a lot of the training samples looking for the objective law.Finally,in the realization of the evaluation model, based on MATALB neural network toolbox established BP network model by real estate price factors as input, real estate price as output, and designed,improved and optimized the network model. Used the true eases by market research to train and evaluate the model validation,achieved the satisfactory results.Keywords:real estate appraisal;BP neural network;price evaluation model;MATLAB...