Research On Prediction Model Of Commercial Residential Housing Price Based On TEI@I Method

Realty industry is one of the most important pillar industry in China. The stable development of the realty industry has a significant impact on the economic growth and the improvement of living standard of the residents. Real estate price is the result of market supply and demand relationship and related factors, which can be taken as the barometer of the market changing. Therefore, establishing a housing price prediction model which is suitable for China’s specific national conditions is very important. The participants of the real estate market can make decesions timely and accurately, which is significantly important for the healthy development of China’s realty industry.At the beginning of the thesis, we summarized the predictiong models of the housing price at home and abroad. Throughout comparative analysis, we got the characteristics of different prediction model. Based on the actual situation of China’s real estate market, we took the TEI@I method as the main method of this paper. And then, we proposed two integrated forecasting model according to the size of the data. The basic train of thought is as follows: predicting the linear sequences by using the econometric models, getting the residual containing of the nonlinear factors of the original sequences, then predicting the residual by using intelligent technology. The final prediction results can be obtained by adding the two prediction results together.In the end, we took the housing price of Shenzhen as a study object and predicted the monthly new commodity housing price based on ARMA and neural network prediction model.At the same time, we also predict the yearly average housing price of the secondary market, based on multiple regression and support vector machine integrated forecasting model. By comparison, we found that the integrated prediction method based on TEI@I method can reduce the prediction error effectively, which has a much more prediction accuracy compared with single prediction model or traditional combined prediction model.