R help - May 2004 - irregular time series

Background:
OS: Linux Mandrake 9.1
release: R 1.9.0 
editor: Xemacs 21.4
frontend: ESS 5.1.23
---------------------------------

Colleagues

I have two time series (upwelling index and water temperature) of evenly
spaced, daily data over 18 months, but the upwelling index  series has a gap
of about 2 months right in the middle of it. I want to do the acf, pacf,
ccf, and a cross-spectral analysis of the two series.

I realise that I could just break each series into two segments and
cross-correlate with the shorter series, but I'd rather deal with the whole
series to increase the nyquist frequency. I think the its function in the
irregular time series package will create a class its object with the right
time stamps, but can this then be used in the same was as a class ts object
for the correlation and spectral anayses?  

Sam
----
Sam McClatchie,
Sub-program leader, Pelagic Fisheries
South Australian Aquatic Sciences Centre
PO Box 120, Henley Beach 5022
Adelaide, South Australia
email <mcclatchie.sam at saugov.sa.gov.au>
Telephone: (61-8) 8200 2448
FAX: (61-8) 8200 2481
Research home page <http://www.smcc.150m.com/>
  
                   /\
      ...>><xX(??> 
                //// \\\\
                   <??)Xx><<
              /////  \\\\\\
                        ><(((??> 
  >><(((??>   ...>><xX(??>O<??)Xx><<

On Thu, May 20, 2004 at 02:23:46PM +0930, McClatchie, Sam (PIRSA-SARDI) wrote:
[long time series, broken in two with a gap]
> I realise that I could just break each series into two segments and
> cross-correlate with the shorter series, but I'd rather deal with the
whole
> series to increase the nyquist frequency. I think the its function in the
> irregular time series package will create a class its object with the right
> time stamps, but can this then be used in the same was as a class ts object
> for the correlation and spectral anayses?  
its does some nice things for storing, plotting, and manipulating
irregular time series, but isn't long on the analysis.

A few options:

1) Analyze the two sub-series separately.
2) There is some merit to de-mean or de-trending both series,
zero-padding the shorter so its length matches the longer, and 
performing spectral analysis that way.  Frequencies near zero
should be treated with suspicion, however.
3) Jim Lindsey has some continuous ARMA and Kalman filter routines
on his site (Google for "Lindsey" and "rmutil", which is the
name of one of those packages).
4) I'm working on an R version of the Lomb periodogram, which
was built for irregular series, but I've no guarantees when I'll
roll it up - rather busy most days.  I do remember seeing an S
version on someone's web page, but that was a while ago, and it
was the "direct" or slow method.

Hope that helps

Jason



-- 
Indigo Industrial Controls Ltd.
http://www.indigoindustrial.co.nz
64-21-343-545
jasont at indigoindustrial.co.nz

The 'its' class may be useful to you for printing/plotting/aligning
data,
but as Jason says, you will need to rely on other packages such as stats
(formerly ts) for analysis, using the appropriate na.action.

Two comments in addition:

Since your data is (from what you say) regular but incomplete, you may well
not need any of the irregular time-series functionality.

In the frequency domain, I should point out that Nyquist frequency is a
reciprocal function of the sample interval (e.g. daily), not the sample
length.  Increasing the sample length will in fact reduce the sampling
interval in the frequency domain, leaving Nyquist unchanged.

- Giles
> -----Original Message-----
> From: r-help-bounces at stat.math.ethz.ch
> [mailto:r-help-bounces at stat.math.ethz.ch]On Behalf Of Jason Turner
> Sent: 20 May 2004 16:45
> To: McClatchie, Sam (PIRSA-SARDI)
> Cc: 'r-help at lists.R-project.org'
> Subject: Re: [R] irregular time series
> 
> 
> On Thu, May 20, 2004 at 02:23:46PM +0930, McClatchie, Sam 
> (PIRSA-SARDI) wrote:
> [long time series, broken in two with a gap]
> > I realise that I could just break each series into two segments and
> > cross-correlate with the shorter series, but I'd rather 
> deal with the whole
> > series to increase the nyquist frequency. I think the its 
> function in the
> > irregular time series package will create a class its 
> object with the right
> > time stamps, but can this then be used in the same was as a 
> class ts object
> > for the correlation and spectral anayses?  
> 
> its does some nice things for storing, plotting, and manipulating
> irregular time series, but isn't long on the analysis.
> 
> A few options:
> 
> 1) Analyze the two sub-series separately.
> 2) There is some merit to de-mean or de-trending both series,
> zero-padding the shorter so its length matches the longer, and 
> performing spectral analysis that way.  Frequencies near zero
> should be treated with suspicion, however.
> 3) Jim Lindsey has some continuous ARMA and Kalman filter routines
> on his site (Google for "Lindsey" and "rmutil", which
is the
> name of one of those packages).
> 4) I'm working on an R version of the Lomb periodogram, which
> was built for irregular series, but I've no guarantees when I'll
> roll it up - rather busy most days.  I do remember seeing an S
> version on someone's web page, but that was a while ago, and it
> was the "direct" or slow method.
> 
> Hope that helps
> 
> Jason
> 
> 
> 
> -- 
> Indigo Industrial Controls Ltd.
> http://www.indigoindustrial.co.nz
> 64-21-343-545
> jasont at indigoindustrial.co.nz
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://www.stat.math.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide! 
http://www.R-project.org/posting-guide.html


********************************************************************** 
This is a commercial communication from Commerzbank AG.\ \ T...{{dropped}}