>>>>> "Nobumoto" == Nobumoto Tomioka <[EMAIL PROTECTED]>
>>>>> on Thu, 27 Feb 2003 15:00:11 -0800 writes:
Nobumoto> Dear Stuff,
[you probably meant "staff" but that's not much better:
There's no paid staff for the R project.
It's all volunteering ! ]
Nobumoto> I am trying to enjoy "PAM", but I could not
Nobumoto> recognize which sample is in which branches in the
Nobumoto> clustering tree because of too many samples and
Nobumoto> size limitation for my monitor.
I first thought you were talking about PAM as in
Kaufman & Rousseeuw (1990) and pam() from the (recommended)
cluster package.
Now I found that you meant the "pamr" CRAN package ...
it would have been helpful if you had told us so.
Also, for packages like these, it is customary to ask the package
maintainers {library(help = pamr) gives you} rather than on R-help.
Further note that when I do
> library(pamr)
I now get
>>
>> Attaching package 'pamr':
>>
>>
>> The following object(s) are masked from package:MASS :
>>
>> enlist
(this is a bit unfortunate, since MASS is a recommended package
in all newer engines of S)
>>
>> There were 48 warnings (use warnings() to see them)
>> > warnings()
>> Warning messages:
>> 1: The use of _ is soon to be removed: you will be warned repeatedly
>> 2: The use of _ is soon to be removed: you will be warned repeatedly
>> 3: The use of _ is soon to be removed: you will be warned repeatedly
>> 4: The use of _ is soon to be removed: you will be warned repeatedly
>> ......
>> ......
{I hope you all take note of this if you haven't before!}
Nobumoto> size limitation for my monitor. I mean, I did
Nobumoto> unsupervised clustering with "PAM" (
Nobumoto> data2$newy<-pamr.makeclasses(data2) )
Nobumoto> with 275 primary samples, but samples were piled up at the
Nobumoto> point of each branch, so I cannot recognize them.
Nobumoto> Can we get the output file with clustered
Nobumoto> information ? Can we change the size of charactor
Nobumoto> ? Please let me know how should I do for it.
Nobumoto> Best regards
I see that the function calls dist() and hclust()
from the "mva" package and also plot.hclust() indirectly.
Also help(pamr.makeclasses) says that the function is fragile.
Unfortunately, it does not allow to pass arguments to plot.hclust()
(it passes everything ("...") only to hclust().
plot(hclust(..), cex = 0.6) would work (if the cex was passed),
but as I see, setting par(cex = 0.6) globally
does not work with plot.hclust().
To solve this, it seems you need to change the
pamr.makeclasses() function, e.g. allowing it to pass a `cex'
parameter.
[BTW: `Netiquette' recommends E-mail signatures of 3-6 lines or so,
*not* the 24 lines you used for all the address information!]
Martin Maechler <[EMAIL PROTECTED]> http://stat.ethz.ch/~maechler/
Seminar fuer Statistik, ETH-Zentrum LEO C16 Leonhardstr. 27
ETH (Federal Inst. Technology) 8092 Zurich SWITZERLAND
phone: x-41-1-632-3408 fax: ...-1228 <><
______________________________________________
[EMAIL PROTECTED] mailing list
http://www.stat.math.ethz.ch/mailman/listinfo/r-help
