[adegenet-forum] dapc technical definitions

[Martin Llewellyn]

Dear Thimbaut,

Thanks very much for the new tutorial. Most helpful for preparing figures. Is there a full list of allowed colours somewhere ? lightpurple, for example, and perhaps for reasons of taste, is not recognised

All the best

M

From: t.jombart at imperial.ac.uk
To: mirainoshojo at gmail.com; adegenet-forum at r-forge.wu-wien.ac.at
Date: Thu, 9 Jun 2011 09:03:34 +0000
Subject: Re: [adegenet-forum] dapc technical definitions










Dear Valeria, 



discriminant functions are principal components of DA. I use the first terminology to prevent ambiguity with PCs of PCA, also used in the method.



Eigenvalues indicate the between/within variance ratio of the corresponding discriminant functions. The first eigenvalue is the largest ratio achievable using linear combinations of PCs of the PCA. The sum of the eigenvalues is merely the sum of these ratio
 computed for all retained variables. It is an indicator of how much between-group variation there is overall in the data, but I've never used this information so far.



Cheers



Thibaut












From: adegenet-forum-bounces at r-forge.wu-wien.ac.at [adegenet-forum-bounces at r-forge.wu-wien.ac.at] on behalf of valeria montano [mirainoshojo at gmail.com]

Sent: 07 June 2011 18:49

To: adegenet-forum at r-forge.wu-wien.ac.at

Subject: [adegenet-forum] dapc technical definitions





Dear Thibaut,



I swear I read the DAPC tutorial, but I couldn't completely clarify myself about a couple of stupid things. If I missed them, feel free to answer "see page xx" and that's it.



First, when I refer to the eigenvalues 1, 2 and so on, do I have to say discriminant "functions" or can I also use the world component? I didn't use this last, but it would be easier to make people absorb them, as it's a new method (anyway if it's wrong
 no probs). Another stupid thing, when referring to the highest eigenvalue, I wrote something like: "this is the value which shows the highest ratio of among/within group variance"...is that the correct definition? (maybe not really elegant). 
Is there any parallel I can do with "the variance explained by this component is x"? Does it make sense to calculate the total of the eigenvalues of the discriminant functions to assign a % of the discrimination ratio they explain?



well, thanks in advance! (sorry for the chaos!)



Valeria






_______________________________________________
adegenet-forum mailing list
adegenet-forum at lists.r-forge.r-project.org
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/adegenet-forum 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/adegenet-forum/attachments/20110609/fbb4f780/attachment.htm>