First thoughts on spatio-temporal classes...

Hi Blair,

I am visiting SAMSI this fall and I would like to get involved in this. Is
there any working group discussing this? If not, perhaps we could have a
chat with all the people interested in the area (samsi, duke, unc, ncsu,...)
and talk about this.

Best wishes,

Virgilio

Sent from my iPhone

On 18/09/2009, at 12:25, "Blair Christian" <blair.christian at gmail.com>
wrote:

Hi all,

This is in response to an earlier post of mine inquiring about options for
spatio-temporal classes in R.

Thank you to all who replied for your interest in spatio-temporal data
classes and methods. ?There were many excellent comments in your emails.
?I
have been attending the SAMSI conference on space-time
analysis<http://www.samsi.info/programs/2009spatialprogram.shtml>this
week and have put together a summary (a working draft) of some

thoughts<http://www.isds.duke.edu/%7Ejbc30/SpatioTemporal/workGroupComp.pdf>I
have about classes.

I also started a bare bones webpage:
http://www.isds.duke.edu/~jbc30/SpatioTemporal/

A also put up the latex source code and bib file I used if you are
interested in making additions, corrections, or other. ?Please email me
any
changes, and I'll maintain an RCS version of it.

I'll put some plots up there when I have time to give you all an idea of
what my context is. ?Ironically, I can't ssh to the webserver from work
(at
the US EPA), so I can only update it from home.


Here are some questions to you all:

- What data do you have interest in analyzing?

- What type of code do you have to contribute?

- Are you interested in writing classes/methods? ?If so, what is your
background? ?(I can provide several tutorials on S4 classes on the web
that
I found useful).

I will pass along these related packages as being related to
spatio-temporal
data:
trip (trip class extends SpatialPointsDataFrame), pastecs, tripEstimation,
hydrosalinity(?), PBSmapping(?), RadioSonde, STARIMA(?), Bayesian Vector
Accumulation(?), spBayes(?), INLA (Havard Rue website)

I skimmed a fair number of time related packages (more than just timeDate
and chron), including xts, zoo, (tseries, timeSeries, ts, its, etc) as
well,
and came to the conclusion that it would best to allow either a time
series
type package (regular or irregular times) as well as a package for
continuous time (functional data, like paths of animals, in the trip
library), and I'm a fan of the fda
library<http://ego.psych.mcgill.ca/misc/fda/>.
It would be nice to have easy spline/FDA access for datasets where time is
collected in a continuous manner rather than in the usual arima grid
setting.

With that information dump, I'll try and get another update out in about
two
weeks depending on my schedule, which should include any progress I've
made
that I'm satisfied with. ?I will respond to individual emails as well.


Cheers,
Blair

? [[alternative HTML version deleted]]