Research On Real Estate Early Warning Based On Artificial Neural Network

After the housing reform in 1998, Chinese real estate industry developed very quickly. However, at the same time, house price rose rapidly and the housing market changed unpredictably, especially in recent years, which certainly will affect the development of the whole industry. It is quite helpful to establish a scientific real estate early-warning system, which could help government, developers and buyers to make more reasonable decisions.There are several methods such as Prosperity Index, Statistics, Model and Artificial Neural Network used in real estate early-warning in existing research. Artificial Neural Network method was used more than others recently, for its lots of features such as nonlinearity, fault tolerance, etc. However, there are still some shortcomings when it was used in recent research.This thesis introduces the idea of building the early-warning system based on artificial neural network. Then, it makes the empirical analysis with a sample of Shanghai, and it chooses ten indicators as warning signs indicators, the growth of sales area and price as alarm indicators. The thesis also collects the quarterly data from 2001 which is related to real estate in Shanghai. Finally, the thesis analyses Shanghai real estate by the early-warning model built on the artificial neural network, the output of the model indicates the real estate of Shanghai in the third quarter of 2010 will be normal.The probable innovation of the thesis concentrates on four aspects:first of all, the thesis chooses quarterly data as the study example. Secondly, it divides the alarm indicators into five levels instead of three levels. Thirdly, it determines the time difference between the warning signs indicators and the alarm indicators by Time Difference Relevance Analysis. Fourthly, it thinks about the influence of government regulation and control policy.