transforming data prior to CCA

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

transforming data prior to CCA

SRuhl
Hi everyone,
I´m a student and relatively new to R so apologies in advance if this question seems stupid or obvious to you.
I have collected a dataset with about 60 species of diatoms (count data from 19 different sample sites) and environmental variables for each site (salinity, pH, etc.). It´s all in the same dataset but distinct in R through the functions below

diat <- diatom [, 1:60]  ##species
envir<- diatom [, 61:66] ##environmental variables

The long-term plan is to perform a canonical correspondence analysis (CCA in the vegan package) on it but the data obviously has to conform to some standarts first. Ideally, any two variables should be in a linear relationship and multivariate normality should be given as well as homoscedasticity (I haven´t tested for this one yet, that´ll be another adventure).  Now my data - surprise - does not conform to a normal distribution nor do the relationships seem linear so I need to transform it (but which parts?). The usual log transformation doesn't change anything so I found this one (the poisson generalized linear model)
glm(formula, family=poisson(link=log), data=envir)

again, it doesn´t work because I dont know what formula to put in.
Any kind of help would be greatly appreciated, I am so lost...
Thanks in advance,
SRuhl


On a side-note: the CCA runs on my data already but what good is that when the data is not in the right format? It may look completely different when the data fits all the requirements.

Reply | Threaded
Open this post in threaded view
|

Fwd: transforming data prior to CCA

Gilson Carvalho
Dear Saskia,

First of all, I think that your question would be better answered in other
list of discussion like [hidden email]. Secondly, CCA does not
assume that your data are normal and homocedastic, the model assume an
unimodal and simetric response of species , but is robust to deviations of
this assumption. But I recomend to you ask this question at the list that I
suggested above.

Best Regards,

Gilson Carvalho


2013/4/5 SRuhl <[hidden email]>

> Hi everyone,
> I´m a student and relatively new to R so apologies in advance if this
> question seems stupid or obvious to you.
> I have collected a dataset with about 60 species of diatoms (count data
> from
> 19 different sample sites) and environmental variables for each site
> (salinity, pH, etc.). It´s all in the same dataset but distinct in R
> through
> the functions below
>
> diat <- diatom [, 1:60]  ##species
> envir<- diatom [, 61:66] ##environmental variables
>
> The long-term plan is to perform a canonical correspondence analysis (CCA
> in
> the vegan package) on it but the data obviously has to conform to some
> standarts first. Ideally, any two variables should be in a linear
> relationship and multivariate normality should be given as well as
> homoscedasticity (I haven´t tested for this one yet, that´ll be another
> adventure).  Now my data - surprise - does not conform to a normal
> distribution nor do the relationships seem linear so I need to transform it
> (but which parts?). The usual log transformation doesn't change anything so
> I found this one (the poisson generalized linear model)
> glm(formula, family=poisson(link=log), data=envir)
>
> again, it doesn´t work because I dont know what formula to put in.
> Any kind of help would be greatly appreciated, I am so lost...
> Thanks in advance,
> SRuhl
>
>
> On a side-note: the CCA runs on my data already but what good is that when
> the data is not in the right format? It may look completely different when
> the data fits all the requirements.
>
>
>
>
>
> --
> View this message in context:
> http://r.789695.n4.nabble.com/transforming-data-prior-to-CCA-tp4663454.html
> Sent from the R help mailing list archive at Nabble.com.
>
> ______________________________________________
> [hidden email] 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.
>


--
Prof. Gilson Correia de Carvalho, M.Sc.
----------------------------------------------------
Pesquisador Associado
Laboratório de Ecologia Bentônica - LEB
Instituto de Biologia - UFBA
Universidade Federal da Bahia - UFBA
----------------------------------------------------
Professor Assistente
Departamento de Biointeração
Instituto de Ciências da Saúde - ICS
Universidade Federal da Bahia - UFBA
-----------------------------------------------------
Diretor Técnico
Holos Soluções Ambientais Ltda
-----------------------------------------------------
Skype: bio_gilson
GTalk: biogilson



--
Prof. Gilson Correia de Carvalho, M.Sc.
----------------------------------------------------
Pesquisador Associado
Laboratório de Ecologia Bentônica - LEB
Instituto de Biologia - UFBA
Universidade Federal da Bahia - UFBA
----------------------------------------------------
Professor Assistente
Departamento de Biointeração
Instituto de Ciências da Saúde - ICS
Universidade Federal da Bahia - UFBA
-----------------------------------------------------
Diretor Técnico
Holos Soluções Ambientais Ltda
-----------------------------------------------------
Skype: bio_gilson
GTalk: biogilson

        [[alternative HTML version deleted]]


______________________________________________
[hidden email] 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.
Reply | Threaded
Open this post in threaded view
|

Re: Fwd: transforming data prior to CCA

SRuhl
Thanks a lot, I will try it :)