The Study Of Early Warning System Of Chinese Real Estate Market Based On Artificial Neural Network

To a certain extent,the stability of the economic development of a country or a region is restricted by the state of the real estate industry development. The rapid development of the real estate industry is likely to cause excessive concentration of funds and resources, in return, the relative shortage of resources in other industries would further undermine the structure of economic development, the arrival of the economic crisis lay hidden. Thus, through the establishment of early warning systems of the real estate, the development of the state in real estate industry can be monitored effectively, avoid abnormal fluctuations in the market, leading the industry in the normal direction and stable development, provided a good foundation for the sustained?healthy and stable development of national economy.Based on collecting and organizing the existing literature, learning and understanding theory of real estate early warning, real estate cycle fluctuation theory and artificial neural networks, respectively selected Shanghai, Hangzhou and Fuzhou as representatives of cities for the first-tier cities, second-tier and third-tier cities, with the help of Chinese real estate statistical yearbook and statistical yearbook of each level cities, collected and collated the relevant real estate market data in the three cities from 2000 to 2015 for empirical research. After analyzed the real estate cycle of the nation and the three cities, combined with the theory of real estate economy,selected 14 indicators as early warning indicators, used the time difference correlation analysis and qualitative analysis method for the selection of the nine indicators as a leading indicator. Based on artificial neural network technology and the use of Matlab software to write and run, built Chinese real estate market early warning models. 14 training set as the input layer, compared a test set of validation output with actual situation, results show that the model can be built for the real estate market early warning of judgment for the next year. The empirical results showed that in 2016 the real estate market in Shanghai and Fuzhou will be in a "normal" state, and in Hangzhou will be "hot" state.Finally, after summarizing research results,the article pointed out existing in the research of alert degree of the sample quantity less, warning degree is not detailedenough and do not include macroscopic policy influence in the model, prospected the future of Real Estate Early Warning Model.