The Prediction of Stock Index Based on Fuzzy Wavelet Analysis

The prediction of financial time series is a puzzle question recognized by all over the world, the traditional methods using linear and complete rational equilibrium pattern have some shortcomings. The paper applies fuzzy set to wavelet analysis and provides a fuzzy wavelet membership function which makes the wavelet coefficients with smaller degree of membership equate zero and the bigger ones shrink toward zero. This kind of de-noising method reflects commendably the impreciseness existing in the system parameters, and thus has better effect of denoising. The method is used for support vector regression model(SVR) and the prediction of Shanghai Stock Exchange Composite Index illustrates its feasibility.