カオス時系列データ予測のための埋込み次元決定方法

Abstract

In a local approximation technique for predicting chaotic time series, a state space is reconstructed from a time series using delay coordinate embedding and then a local predictor is constructed in the state space. The embedding dimension of the state space has a significant effect on the prediction accuracy. We evaluate the prediction error as a function of the embedding dimension for chaotic time series generated by the Hénon map and the Ikeda map and show basic guidelines for determining the embedding dimension.