[R] R code for estimating Hurst exponent

[Martin Maechler]

>>>>> "CathW" == Xiaozhe Wang <Catherine.Wang at infotech.monash.edu.au>
>>>>>     on Fri, 05 Dec 2003 10:52:05 +1100 writes:

    CathW> Has anyone writen R code for estimating Hurst exponent with R/S method 
    CathW> or other methods?
    CathW> or any other source of R code available?

There is the  fracdiff  package on CRAN for fitting

frARIMA(p,d,q) models with fractional "d" and there's a
one_to_one relationship between  'd' and the Hurst parameter for
these models: d = H - 1/2.

The R/S method is known to be far from optimal for many years
now, and there are better methods for which I'd recommend at
least the book 

@Book{BerJ94,
  author =       {Jan Beran},
  title =        {Statistics for Long-Memory Processes},
  publisher =    {Chapman \& Hall},
  year =         1994,
  series =       {Monographs on Statistics and Applied Probability 61},
  address =      NY
}

--------

In the book appendix, he has S-plus code (functions and a
script, not all really ok, using global variables, ...)  that I
had started to package for R many months ago, package name
"longmemo".

The version was never close to fulfill any CRAN quality
standards on R packages --- though I can make it available if
you are interested; it's already quite a bit more useful than
the original fortran code {I'm also sending this (BCC) to Prof.Beran
to solicit comments}. 

Martin Maechler <maechler at stat.math.ethz.ch>	http://stat.ethz.ch/~maechler/
Seminar fuer Statistik, ETH-Zentrum  LEO C16	Leonhardstr. 27
ETH (Federal Inst. Technology)	8092 Zurich	SWITZERLAND
phone: x-41-1-632-3408		fax: ...-1228			<><