On a Model Approach to Forecasting of Chaotic Time Series. II

Based on a previous work of the present author, a forecasting method for chaotic time series is proposed and applied to some artificial time series generated by deterministic chaos equations. It is shown that the efficiency of the method depends on the coarseness of the discrete information space and the length of the model time series that is used to construct the forecasting function. In a previous paper, 1) the forecasting of chaotic time series of a certain type generated by deterministic equations was studied by the present author. In those cases, the n-n + 1 plots, the relation between the observed value of the variable of the time series at the nth step and that at the (n + 1)th step, takes the form o