Wavelet Neural Network And Its Application To Forecasting For Residential Buildings Market

Wavelet analysis is a newly developed theory, which overcomes the disadvantages of traditional Fourier analysis. It has superior localized features in both the times and the frequencies domain and can be applied to signal processing, image processing and speech analysisThe paper gives a description of the basic theory of economic forecasting ,makes a research on the linear regression model ,gray forecasting model and bp neural network in the economic forecasting,and introduces combined-forecasting model to optimized the three forecasting model above.It is shown that the combined-forecasting model.It can efficiently avoid losting the important economic information,and gain the better result than the three forecasting model above.Finally the paper studies on the wavelet neural network applications on prediction of residential buildings market.It overcome the difficulties and draw backs in modeling non-stationary time serise by using linear forecasting.It apply the wavelet neural network ,combined one parameters method ,to predict the real estate market in the future.Dased on the wavelet analysis,wavelet neural network ,whick has neural network idea,uses wavelet function represent sigmoid function in hided-layer.So it can reset shape-parameter of wavelet and weight in the procession of optimizing error function.In this paper corresponding learning algorithm is put forward to calculate the dilation and shifts which are the parameters of wavelet function and the weight of neural network structure.After preprocessed ,samples are input to network.The trainned wavelet neural network reduces epochs ,recreases forecasting accurate and has good abilities of function learning.The method has strong universalized capabilities and the abilities to adapt to the changes of function also.