[R] forecasting from fracdiff objects

Martin Maechler maechler at stat.math.ethz.ch
Mon Jun 14 18:59:20 CEST 2004

>>>>> "Alan" == Alan Simpson <simpsa02 at maths.uwa.edu.au>
>>>>>     on Mon, 14 Jun 2004 12:36:58 +0800 writes:

    Alan> Does anybody know if it is possible to forcast or predict from a 
    Alan> fracdiff object?

yes, it is possible.... but not (yet?) with a builtin function,
but rather by writing code to do it --- using the fracdiff()
result *and* the data from which you want to make h-ahead
predictions (which I assume is what you want).

At the moment, there's not even  residuals() or fitted()
methods available for fracdiff objects.
At least those are really just an exercise to implement
(which might not be true for the h-ahead predictions:
 There you'd probably need a version of the Durbin-Lewinson or the
 innovations algorithm).

User contributions are very welcome... ;-)

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			<><

More information about the R-help mailing list