LongMemoryModelling {fSeries}R Documentation

Long Memory Behaviour of Time Series

Description

A collection and description of functions to investigate the long memory behavior of an univariate time series process.

The functions and methods are:

fgnSim Simulates Fractional Gaussian Noise.

Sorry, not yet implemented are the functions to fit the Hurst exponent.

Usage

fgnSim(n = 1000, H = 0.7, method = c("beran", "durbin", "paxson"), 
mean = 0, std = 1)

Arguments

H the Hurst exponent, a numeric value between 0.5 and 1, by default 0.7.
mean, std mean and standard deviation of the random innovations. By default a zero mean and a unit standard deviation is assumed.
method the method how to generate the time series sequence, one of the following character strings: "beran", "durbin", or "paxson".
n number of data points to be simulated, a numeric value, by default 1000.

Value

returns a numeric vector of length n, the fractional Gaussian noise series.

References

Paxson V. (1995); Fast Approximation of Self-Similar Network Traffic, Berkeley.

Examples

## fgnSim -
   par(mfrow = c(3, 1), cex = 0.75)  
   # Beran's Method:
   plot(fgnSim(n = 200, H = 0.75), type = "l",  
         ylim = c(-3, 3), xlab = "time", ylab = "x(t)", main = "Beran")
   # Durbin's Method:
   plot(fgnSim(n = 200, H = 0.75, method = "durbin"), type = "l",
         ylim = c(-3, 3), xlab = "time", ylab = "x(t)", main = "Durbin")
   # Paxson's Method:
   plot(fgnSim(n = 200, H = 0.75, method = "paxson"), type = "l",
     ylim = c(-3, 3), xlab = "time", ylab = "x(t)", main = "Paxson")

[Package fSeries version 200.10058 Index]