Chaotic Time Series Analysis

Provides several algorithms for the purpose of detecting chaotic signals inside univariate time series. We focus on methods derived from chaos theory which estimate the complexity of a dataset through exploring the structure of the attractor. We have taken into account the Lyapunov exponents as an ergodic measure. We have implemented the Jacobian method by a fit through neural networks in order to estimate both the largest and the spectrum of Lyapunov exponents. We have considered the full sample and three different methods of subsampling by blocks (non-overlapping, equally spaced and bootstrap) to estimate them. In addition, it is possible to make inference about them and know if the estimated Lyapunov exponents values are or not statistically significant. This library can be used with time series whose time-lapse is fixed or variable. That is, it considers time series whose observations are sampled at fixed or variable time intervals. For a review see David Ruelle and Floris Takens (1971) , Ramazan Gencay and W. Davis Dechert (1992) , Jean-Pierre Eckmann and David Ruelle (1995) , Mototsugu Shintani and Oliver Linton (2004) , Jeremy P. Huke and David S. Broomhead (2007) .


Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.


0.1-2 by Julio E. Sandubete, 20 days ago

Browse source code at

Authors: Julio E. Sandubete [aut, cre] , Lorenzo Escot [aut]

Documentation:   PDF Manual  

Task views: Time Series Analysis

GPL (>= 2) license

Imports xts, zoo, outliers, nnet, pracma, sandwich, NeuralNetTools

See at CRAN