Q: Does GRASS do Canonical Variate Analysis?

[Agustin Lobo]

Regardless whether Grass does can. analysis or  not, I think that the best
is to:

(i) define your training areas in grass.
(This could be  the worst part, as the rgb display of imagery in grass
is very poor)
(ii) extract the data with r.stats and one mask for each training area.
(iii) pass the data to a Stats package and calculate the transformation
matrix.
(iv) apply the transf. matrix with an script based on r.mapcalc.

In this way you'll have all the possibilities that a good Stat package
would give you for the many "nuances" of multivariant analysis. I suggest
using R, which is also public domain.

Agus

Dr. Agustin Lobo
Instituto de Ciencias de la Tierra (CSIC)
Lluis Sole Sabaris s/n
08028 Barcelona SPAIN
tel 34 93409 5410
fax 34 93411 0012
alobo at ija.csic.es
http://pangea.ija.csic.es/alobo


On Wed, 28 Jul 1999, Terry Duell wrote:

> Hullo all,
> We would like to examine Canonical Variate Analysis as an approach to
> image analysis.
> We currently use GRASS 4.2 (along with other tools), but have not
> upgraded GRASS for a while.
> We are aware that quite a bit has been happening on the GRASS scene,
> and wondered if anyone can advise if the latest versions of GRASS can
> support this form of analysis.
> 
> Thanks for your help.
> 
> Cheers,
> 
> --
> Terry Duell, Senior Mobility Engineer
> Army Engineering Agency
> Maribyrnong, Victoria, Australia
> ph:61-3-93195837 fax:61-3-93195830
> 
> 
> 
>