A SMOOTHING ALGORITHM FOR NONLINEAR TIME SERIES
Steve A. Billings, Kian L. Lee · International Journal of Bifurcation and Chaos · 2004
A new NARMA based smoothing algorithm is introduced for chaotic and nonchaotic time series. The new algorithm employs a cross-validation method to determine the smoother structure, requires very little user interaction, and can be combined with wavelet thresholding to further enhance the noise reduction. Numerical examples are included to illustrate the application of the new algorithm.