[R] what is scaling in plot

[Gavin Simpson]

In an ordination such as this, you have two sets of scores; i) one
pertaining to the (dis)similarity of your samples in terms of species
composition, and ii) one pertaining to the species and which species
co-occur. You can't display both sets of information and retain the
meaning of the scores on the one plot. Scaling controls what relationships
are preserved (focussing on sites or species etc). Gabriel (forget the
citation now but is was in Biometrika) suggests that scaling 3 (symmetric
scaling) is optimal and to be preferred.

There is a discussion of the various scalings in the Design Decision
vignette that you should read. If this is too complex, then perhaps start
with Jan Leps and Petr Smilauer's book "Multivariate analysis of
ecological data using Canoco", 2003, Cambridge University Press.

HTH

Gavin

> code
>
> rm(list=ls())
> library(vegan)
> library(MASS)
>
> data(varespec)
> print(varespec)
> str(varespec)
>
>
> #PCA
> vare.pca  <-  rda(varespec)
> vare.pca
> plot(vare.pca)
> sum(apply(varespec, 2, var))
> biplot(vare.pca,  scaling  =  -1)
>
> Elaine
>
> On Sat, Aug 14, 2010 at 10:48 AM, Ben Bolker <bbolker at gmail.com> wrote:
>
>> elaine kuo <elaine.kuo.tw <at> gmail.com> writes:
>>
>> > Pls kindly advise what scaling is in plot.
>> > Sometime it could be negative but sometimes it might be positive
>> > .(I guess it is the proportion between the plot and the margin)
>>
>>   Your question is unclear. Please give more context and/or details.
>>
>> ______________________________________________
>> R-help at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-help
>> PLEASE do read the posting guide
>> http://www.R-project.org/posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.
>>
>
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>