streamChaos - Analysis of Nonlinear Data Streams - R package

A framework for analyzing nonlinear data streams (and nonlinear time series).

Based on the package:

stream

Installation

In development.

Usage

Load the package and create a data stream based on a transient Logistic Map:

$x_t = r_t \cdot x_{t-1} (1 - x_{t-1})$,

in which $r_t \in [2.0, 4.0]$.



In [2]:

    
options(warn=-1)

library("streamChaos")

dsd <- NLDSD_TransientLogisticMap(N=12000)

Process such stream using the Permutation Entropy algorithm with a sliding window of $800$ observations, sliding $10$ observations by iteration.



In [3]:

    
pe <- DSCDD_PermutationEntropy(m=5, d=1, window.length=800)

ret <- processStream(dsd, pe, window.step=10)

Plot the results.



In [10]:

    
options(repr.plot.width=7, repr.plot.height=5)

plot(ret$ds[,1], pch='.', ylab='x')
par(new=T, xaxt='n', yaxt='n')
ts.plot(ret$measures, col=2, xlab='', ylab='', lwd=2)

References

Algorithms:

[Permutation Entropy]
[Recurrence Quantification Analysis]
[Multidimensional Fourier Transform]
[Permutation-Invariant]

Data streams:

[Transient Logistic Map]
[Transient Lorenz Attractor]