[R] using deltat parameter in time series in HoltWinters prediction

John jwd at surewest.net
Mon Jan 9 02:03:27 CET 2012


On Sat, 7 Jan 2012 14:54:30 +0100
Antonio Tirri <antonio.tirri at gmail.com> wrote:

> On 7 January 2012 14:34, Prof Brian Ripley <ripley at stats.ox.ac.uk>
> wrote:
> 
> >
> > And what did your statistics advisor say?  The problem is your
> > understanding of the words 'sampling period'.
> >
> 
> I don't have a statistics advisor, my computer networks tutor said to
> me "try to forecast time series values with ARIMA models".
> And i saw R is a powerful tool for this target.
> The only information I have about this time series (
> http://www.grid.unina.it/Traffic/Traces/Magnets.php) is that data are
> sampled every 0.05 seconds, so i'm trying to understand the meaning
> of the words "sampling period" since you said my suppositions are not
> true.
> 
Given that your sampling interval is 0.05 seconds, you are acquiring
more than 1.7 million data points in 24 hours.  A weeks worth of data
are going to be on the order of 12 million data points.  Are you
attempting and hourly, daily or weekly forecast?

JWD



More information about the R-help mailing list