I make quite a lot of use of irregular time-series, and had already spent a
bit of time writing an 'its' class when the 'irts' class was
released via
the package 'tseries'.
I have experimented with the 'irts' class, and have some practical
issues
with its use. In some applications of irregular time-series (in my case
these are financial and econometric) there are many instances where it is
useful to consider them as an extension (subclass) of the class
'matrix',
using S4 class inheritance. This is also useful in printing.
The 'its' class I have written is an S4 extension of the matrix class,
using
a single slot for a POSIX dates vector, which time-indexes the rows. The
practical advantage of this, compared to the 'irts' class, is that the
matrix methods are inherited.
In addition to file operations and some simple functions, I have also
written a number of method extensions for this 'its' class, including
[.its,
[[.its, intersect, union, diff, lag.
Is there value in such a package, given the resemblance of 'its' to the
existing 'irts' class? I would be happy to contribute it, and provide
some
maintenance.
Giles Heywood
Senior Quantitative Analyst
Commerzbank Securities
**********************************************************************
This is a commercial communication from Commerzbank AG.\ \ This ... {{dropped}}
On Fri, Jun 06, 2003 at 09:46:04AM +0100, Heywood, Giles wrote:> I make quite a lot of use of irregular time-series, and had already spent a > bit of time writing an 'its' class when the 'irts' class was released via > the package 'tseries'. > > I have experimented with the 'irts' class, and have some practical issues > with its use. In some applications of irregular time-series (in my case > these are financial and econometric) there are many instances where it is > useful to consider them as an extension (subclass) of the class 'matrix', > using S4 class inheritance. This is also useful in printing. > > The 'its' class I have written is an S4 extension of the matrix class, using > a single slot for a POSIX dates vector, which time-indexes the rows. The > practical advantage of this, compared to the 'irts' class, is that the > matrix methods are inherited. > > In addition to file operations and some simple functions, I have also > written a number of method extensions for this 'its' class, including [.its, > [[.its, intersect, union, diff, lag. > > Is there value in such a package, given the resemblance of 'its' to the > existing 'irts' class? I would be happy to contribute it, and provide some > maintenance.Yes, please! Either inside tseries, assuming Adrian is happy with that, or outside. I use the 'padding irregular series with NAs' trick quite a bit and would love to have its support. Dirk -- Don't drink and derive. Alcohol and analysis don't mix.
Heywood, Giles wrote:>I make quite a lot of use of irregular time-series, and had already spent a >bit of time writing an 'its' class when the 'irts' class was released via >the package 'tseries'. > >I have experimented with the 'irts' class, and have some practical issues >with its use. In some applications of irregular time-series (in my case >these are financial and econometric) there are many instances where it is >useful to consider them as an extension (subclass) of the class 'matrix', >using S4 class inheritance. This is also useful in printing. > >The 'its' class I have written is an S4 extension of the matrix class, using >a single slot for a POSIX dates vector, which time-indexes the rows. The >practical advantage of this, compared to the 'irts' class, is that the >matrix methods are inherited. > >In addition to file operations and some simple functions, I have also >written a number of method extensions for this 'its' class, including [.its, >[[.its, intersect, union, diff, lag. > >Is there value in such a package, given the resemblance of 'its' to the >existing 'irts' class? I would be happy to contribute it, and provide some >maintenance. >Yes, I think so. There seem to be different ways to do these things and some work better than others, often for reasons that are not obvious in the beginning.> >Giles Heywood >Senior Quantitative Analyst >Commerzbank Securities > > >********************************************************************** >This is a commercial communication from Commerzbank AG.\ \ This ... {{dropped}} > >______________________________________________ >R-devel@stat.math.ethz.ch mailing list >https://www.stat.math.ethz.ch/mailman/listinfo/r-devel > >
"Heywood, Giles" wrote:> I make quite a lot of use of irregular time-series, and had already spent a > bit of time writing an 'its' class when the 'irts' class was released via > the package 'tseries'. > > I have experimented with the 'irts' class, and have some practical issues > with its use. In some applications of irregular time-series (in my case > these are financial and econometric) there are many instances where it is > useful to consider them as an extension (subclass) of the class 'matrix', > using S4 class inheritance. This is also useful in printing. > > The 'its' class I have written is an S4 extension of the matrix class, using > a single slot for a POSIX dates vector, which time-indexes the rows. The > practical advantage of this, compared to the 'irts' class, is that the > matrix methods are inherited. > > In addition to file operations and some simple functions, I have also > written a number of method extensions for this 'its' class, including [.its, > [[.its, intersect, union, diff, lag. > > Is there value in such a package, given the resemblance of 'its' to the > existing 'irts' class? I would be happy to contribute it, and provide some > maintenance.(First sorry for the late reply). Yes of course. My suggestion is that you provide us with its (as a standalone package, or I would be happy to include it in tseries), and we can all experiment a bit with both classes and maybe merge them later on?> > Giles Heywood > Senior Quantitative Analyst > Commerzbank Securitiesbest Adrian -- Dr. Adrian Trapletti Phone : +41 (0) 1 994 5631 Trapletti Statistical Computing Mobile: +41 (0)76 370 5631 Wildsbergstrasse 31 Fax : +41 (0) 1 994 5631 CH-8610 Uster Email : mailto:a.trapletti@bluewin.ch Switzerland WWW : http://trapletti.homelinux.com