Forecast entropy
W. Yao
(Department of Applied Mathematics, University of Western Ontario, London, Ontario, Canada)
C. Essex
(Department of Applied Mathematics, University of Western Ontario, London, Ontario, Canada)
P. Yu
(Department of Applied Mathematics, University of Western Ontario, London, Ontario, Canada)
M. Davison
(Department of Applied Mathematics, University of Western Ontario, London, Ontario, Canada)
575
Abstract
A technique, called forecast entropy, is proposed to measure the difficulty of forecasting data from an observed time series. When the series is chaotic, this technique can also determine the delay and embedding dimension used in reconstructing an attractor. An ideal random system is defined. An observed time series from the Lorenz system is used to show the results.
Keywords
Citation
Yao, W., Essex, C., Yu, P. and Davison, M. (2004), "Forecast entropy", Kybernetes, Vol. 33 No. 5/6, pp. 1009-1015. https://doi.org/10.1108/03684920410534056
Publisher
:Emerald Group Publishing Limited
Copyright © 2004, Emerald Group Publishing Limited