Application Of Artificial Neural Networks To The Real Estate Price Predicting

During30years of construction and development, China's continuously expanding real estate industry has become the national economy's backbone industry after a series of marketization reform such as residential currency and commercialization, management system for the paid use of land, etc. The high risk and high yield of real estate industry as bring the increasingly serious speculation and profiteering. Meanwhile, real estate is capital-intensive industry. Therefore, how to fully grasp the information resources, effectively predict the real estate market, and make the government and real estate enterprises accurately predict and judge the of real estate market, has become the important and urgent research subject with a strong theoretical and realistic significance.The common methods of real estate market price forecast are linear regression forecasting model, time series prediction, the trend line forecast method, etc. Meanwhile, the main data indexes influencing the real estate price involve many uncertain factors, and the relationships between the factors are complex. Actually real estate price predicting is a nonlinear problem. Artificial neural network modeling based on inner link of data itself, overcomes traditional linear analysis and personnel subjective judgment. It has good effect of self-organization, adaptability, also has strong learning ability and the anti-interference ability.Based on artificial neural network as the main research tools, first of all, the real estate price characteristics and its influencing factors are summarized. Through analysis we can find that the residential housing price are sensitive to GDP, the per capita disposable income level, the consumer price index, land development investment, real estate investment factors, etc. Secondly, an artificial neural network theory and methods are introduced, emphatically the BP network model, the PCA-BP network model and Elman network model. Finally, combined with artificial neural network theory and methods and real estate price affecting factors, taking residential housing prices in Yangzhou city for example, the commercial real estate price prediction models are established. Then train the introduced network models, predict with trained networks, make comparisons between the forecast results and the real results, finally summarize the method advantages and improvements. The end of the article put forward the shortcomings of this paper and subsequent research prospect.