Analyses of Optimal Embedding Dimension and Delay for LocalLinear Prediction Model

Abstract

In the reconstructed phase space, a novel local linear prediction modelis proposed to predict chaotic time series. The parameters of theproposed model take the values that are different from those of thephase space reconstruction. We propose a criterion based on predictionerror to determine the optimal parameters of the proposed model. Thesimulation results show that the proposed model can effectively makeone-step and multi-step prediction for chaotic time series, and theone-step and multi-step prediction accuracy of the proposed model issuperior to that of the traditional local linear prediction.