The Application of Takens Theory and Wavelets to Nonlinear Prediction
Zhou Bai · 2001
As a new kind of tool, wavelet plays a more and more important role in all kinds of fields; time series analysis is the basis of prediction, in which a lot of current methods are still used in many ways. Wavelet is combined with time series analysis on the basis of Takens theory in order to realize nonlinear prediction of time series. Firstly, we have made long term nonlinear prediction of a chaos attractor and short term nonlinear prediction for economic time series; Furthermore, we have suggested that we could make the error small enough to be omitted if we use sufficient wavelet basis functions φ mn to approximate to original time series.