Chapter from Time Series Prediction: Forecasting the Future and Understanding the Past, Weigend, A.S. and Gershenfeld, N.A. (ed.), SFI Studies in the Sciences of Complexity, Proc. Vol XV, Addison-Wesley, 1994
Abstract
This chapter compares the success of several nonlinear prediction techniques applied to Data Set A of the Santa Fe Time Series Prediction and Analysis Competition (both A.dat and A.cont). The advantages of a new approach making predictions based on selective use of several different delay reconstructions are illustrated, and a comparison of both local linear and local nonlinear predictions is given. In addition, the phase coherence of the system and the self-consistency of the data is examined using the longer data set A.cont; the latter locates a possible sensor failure in this data set. Limitations due to the amount of data, the sampling rate, and the saturation in the data, in combination with the quality of the predictions achieved with very little information on the value of the initial condition (32 bits of less), suggest that, while the system is clearly nonlinear, evidence from A.dat for sensitive dependence on initial condition, if any, is slight.