-------- Original Message --------
Subject: Re: query [DA with discrete characters - the mod]
Date: Wed, 7 Oct 2009 06:05:58 -0700 (PDT)
From: [email protected]
To: [email protected]
References: <[email protected]>


A "dirty trick" for carrying out discriminant analysis on non-continuous
characters is Anderson's Canonical Analysis of Principal Coordinates
(CAP). The idea is simply to run an initial Principal Coordinates
Analysis, which can use any distance measure appropriate for your type of
data, and then do a traditional discriminant analysis/CVA on the principal
coordinate scores for the "first few" PCoA axes.

You can do this in any software that includes PCoA and DA/CVA (e.g. R or
Past), or you can use Anderson's special-purpose software "CAP" that
includes a number of useful features:

http://www.stat.auckland.ac.nz/~mja/Programs.htm

I am not going to comment on whether CAP is a valid procedure or not, so
don't shoot at me, but it does produce good-looking plots  :-)


Oyvind Hammer
Natural History Museum
University of Oslo


-------- Original Message --------
Subject: query
Date: Wed, 7 Oct 2009 02:45:52 -0700 (PDT)
From: Gabriele Salvo <[email protected]>
To: [email protected]


Hi all, I'm new to the field and I would like to receive a few
suggestions. I have compiled a morphological character/species matrix
including only discrete characters (presence/absence and multistate),
which I would like to analyze in a phylogenetic framework. Since in my
group of interest there is a species complex, I would like to use the
relevant bits of the morphological matrix to visualize the species
complex in morphological space. I have in mind a PCA kind of plot
showing the different entities of the species complex with circles. Some
will overlap completely, others less. I would also like to find out
which are the most useful morphological characters for teasing the
entities of the species complex apart. I have read some articles on
traditional morphometrics and it appears that what I intend to do comes
closer to a Discriminant Analysis, but I learn that this is mainly
(only?) for continuous characters.
Do you have any suggestions (also regarding the software that I could
use)?
Cheers
gabriele



----------------------------------------------------------------------------------------------------------------------------------------------------
Gabriele Salvo
PhD Student
Institut für Systematische Botanik
Universität Zürich
Zollikerstrasse 107
CH-8008 Zürich
SWITZERLAND
e-mail: [email protected]
tel: 0041 (0)44 634 84 18




--
Replies will be sent to the list.
For more information visit http://www.morphometrics.org






--
Replies will be sent to the list.
For more information visit http://www.morphometrics.org

Reply via email to