Abstract: Housing market is a complex system, and housing prices are results of various factors. The prediction accuracy of traditional single forecasting model is not enough to well support economic decision-making. Based on the TEI@I idea, an integrated prediction model is created to predict the direction and level of change in the real estate market, using key predictive indicators based on the dynamic circulation system of the real estate market. Firstly, the changing direction of the real estate market is predicted by boom analysis. Then, based on the divide-and-rule idea, integrated data characteristic driven forecasting model of the real estate market is established. The model quantitatively predicts the real estate market and demonstrates the effectiveness of different forecasting methods by comparing data decomposition and data feature-driven basic model selection. This paper concludes that applying the appropriate base model according to the data feature can predict the investment, demand and price of the real estate market more accurately than the single prediction model. This research enriches the theory and method of real estate market forecasting and predicts the trend of the real estate market more accurately. Moreover, it provides recommendations for the government to design policies and make decisions, for real estate developers to invest and for residents to purchase houses.

Key words: complex system, data characteristic driven, real estate market, integrated forecasting, TEI@I