Dear Friends,
Here are some replies I received on the above subject.
Thank you for the help.
Sincerely,
Alexandre
Dear Dr. Souza,
I was advised to use regression trees, so you have solved one problem
for me, what is the name of the package!
For your question, I can suggest two ways which might work, based on
their use in displaying results of non-metric multidimensional scaling
(metaMDS in R):
meta.x<-metaMDS(x)
plot(meta.x, display="sites", type="t")
Here, the "display" command says to plot sites rather than species, and
the "type" command specifies "text" (which should identify the plot)
rather than "points" which would just put a symbol.
Another possible solution is the"identify" command:
plot ( x, "sites")
identify (plot ( x,"sites"), "sites")
Using this command, you point to each symbol on the chart and click -
causing the name of that plot
to appear., but the result is not saved in the graphic by R until you
have clicked o all of the points.
I hope that this is helpful.
I found the attached tutorial helpful. I believe that it addresses the
question you posed regarding alternate distance matrices. Let me know
if you have any further questions.
http://www.ualberta.ca/~jspence/Spence_lab/MRT-r-package.html
This tutorial presents how to actually implement db-MRT.
Here is the information that you need to see species names (my previous
message was probably of no use).
Download the documentation for package (mvpart) at:
http://bm2.genes.nig.ac.jp/RGM2/R_current/library/mvpart/man/mvpart.html
In it you will find the different arguments with a brief comment about
each, also an example, which is the most useful.
These lines in the example produces two graphics, both of which identify
and locate species. To see both of them open the graphics window in the
R window (if you are using Windows) click on "History" in the Menu bar
and click on "Recording". Then you can use Previous and Next in that
same menu.
mvpart(data.matrix(spider[,1:12])~herbs+reft+moss+sand+twigs+water,spider,xv="p")
# pick the tree size
# pick cv size and do PCA
fit <-
mvpart(data.matrix(spider[,1:12])~herbs+reft+moss+sand+twigs+water,spider,xv="1se",pca=TRUE)
rpart.pca(fit,interact=TRUE,wgt.ave=TRUE) # interactive PCA plot of
saved multivariate tree
I modified he first line as follows
mvpart(data.matrix(spider[,1:12])~herbs+reft+moss+twigs+water,spider,xadj=2,
yadj=3,xv="min",pca=T)
, using xadjand yadj to increase the size of the bar plots. I do not
know how to move the legend out of the way of the tree diagram, though.
Hi Alexandre, I don't think you can do it with mvpart since mvpart
doesn't take the dissimilarity structure into account.
There are other regression tree packages though, and you might want to
look at randomForest which generally provides better results than
classic CART analysis.
In package&function 'rpart' there is a switch to select the
dissimilarity measure.
Depending on your dataset you may be able to change the dissimilarity
measure by simply transforming the data, for example, a sqrt transform
would give square-chord dissimilarity from euclidian distance.
Dr. Alexandre F. Souza
Programa de Pós-Graduação em Biologia: Diversidade e Manejo da Vida
Silvestre
Universidade do Vale do Rio dos Sinos (UNISINOS)
Av. UNISINOS 950 - C.P. 275, São Leopoldo 93022-000, RS - Brasil
Telefone: (051)3590-8477 ramal 1263
Skype: alexfadigas
[email protected]
http://www.unisinos.br/laboratorios/lecopop
--
Esta mensagem foi verificada pelo sistema de antiv�rus e
acredita-se estar livre de perigo.
