----- Forwarded message from Paolo Piras
<paolo.pi...@uniroma3.it> -----
Date: Fri, 10 May 2013
04:36:26 -0400
From: Paolo Piras <paolo.pi...@uniroma3.it>
Reply-To: Paolo Piras <paolo.pi...@uniroma3.it>
Subject: RE:
Cluster analysis
To: "morphmet@morphometrics.org"
<morphmet@morphometrics.org>
phangorn package (and other tons of packages) do that in R
I wrote a function that takes a matrix and a factor and performs cluster analysis (both UPGMA and NJ are displayed) using per-level means.
clusteronmeans<-function(x,group){
library(MASS)
library(car)
library(phangorn)
group2<-group[order(group)]
x2<-x[order(group),]
x3<-aggregate(x2,by=list(group2),FUN=mean)
rownames(x3)<-levels(group2)
x11()
plot(upgma(as.matrix(dist(x3[,-1],method="euclidean"))))
par(new=T)
title(main="upgma")
x11()
plot(nj(as.matrix(dist(x3[,-1],method="euclidean"))))
par(new=T)
title(main="neighbor-joining")
}
let me know if this helps
best
paolo
Inviato: venerdì 10 maggio 2013 10.23
A: morphmet@morphometrics.org
Oggetto: Re: Cluster analysis
----- Forwarded message from Saber Sadeghi <sabersade...@yahoo.com>
-----
Date: Thu, 9 May 2013 03:13:00 -0400
From: Saber Sadeghi <sabersade...@yahoo.com>
Reply-To: Saber Sadeghi <sabersade...@yahoo.com>
Subject: Re: Cluster analysis
To: morphmet@morphometrics.org
|
Dear Dan
NTSYSpc is a good software for UPGMA, but it's not free.
success
saber
--- On Fri, 5/3/13, morphmet_modera...@morphometrics.org <morphmet_modera...@morphometrics.org> wrote:
|
----- End forwarded message -----
----- End forwarded message -----
