Online Time Series Forecasting Based on Biorthogonal Wavelet Kernel Support Vector Machine

As a Special wavelet, biorthogonal wavelet has many advantages in signal processing. This paper constructs a new biorthogonal wavelet based on CDF method and constructs the biorthogonal wavelet kernel function. Then we study the update of incremental model and propose online forecasting algorithm. We research the algorithm based on biorthogonal wavelet kernel support vector machine (SVM) and use this algorithm to forecast the Chinese Csi 300 stock index futures at last. The experiment result shows the algorithm has ideal prediction effect.