Hi Changbin,
I looked at your code again, and it appears as if you're using the mapping
plot for something that it isn't meant for. The mapping shows you how many
points you have in every circle, and these points are represented by the
labels. Your first plot gives the majority vote.
This said, you can hack the function using:
classif <- predict(nir.xyf)
tmp <- table(classif$unit.classif,classif$prediction)
label <- colnames(tmp)
label <- apply(tmp!=0,1,function(x){label[x]})[classif$unit.classif]
label[-match(1:16,classif$unit.classif)] <- ""
cl <- colors()
bgcols <- rev(heat.colors(4))
plot(nir.xyf,
type="mapping",bgcol=bgcols[as.numeric(as.factor(temp.predict))],
main="Mapping plot",labels=label)
It does not calculate the majority vote itself, it just assigns a label to
the category based on the predicted labels. Which is equivalent in this
case.
Cheers
Joris
On Wed, Jun 2, 2010 at 5:08 AM, Changbin Du <changb...@gmail.com> wrote:
>
> library("kohonen")
> data(nir)
> attach(nir)
>
> #SOM, the supervised learning, train the map using temperature as the class
> variable.
> set.seed(13)
> nir.xyf<- xyf(data=spectra, Y=classvec2classmat(temperature), xweight =
> 0.9, grid=somgrid(4, 4, "hexagonal"))
>
>
> temp.xyf <- predict(nir.xyf)$unit.prediction #get prediction
> temp.predict<- as.numeric(classmat2classvec(temp.xyf)) #change matrix to
> vectors.
>
> par(mfrow=c(1,2))
>
> plot(nir.xyf, type="property", property=temp.predict, palette.name=rainbow,
> main="Prediction ")
>
>
> cl <- colors()
> bgcols <- cl[2:14]
> plot(nir.xyf, type="mapping", labels=nir$temperature,
> bgcol=bgcols[as.integer(temp.predict)],
> main="Mapping plot")
>
> par(mfrow=c(1,1))
>
>
>
> HI, Joris,
>
> Thanks so much for your suggestion! I have modified the above codes, and
> what I want is to label the notes by the temperature.
> if a note has 3 objects mapped to it (the temperature are 30, 40, 30), then
> I want the 30 be labeled on the note.
>
> the right plot is the mapping plot, I want it to be labeled by only one
> temperature.
>
> Thanks so much!
>
>
>
>
>
>
>
>
>
>
>
> On Tue, Jun 1, 2010 at 5:36 PM, Joris Meys <jorism...@gmail.com> wrote:
>
>> Dear Changbin,
>>
>> Please provide a self-contained, minimal example, meaning the whole code
>> should run and create the plot as it is now, without having to load your
>> dataset (which we don't have). Otherwise it's impossible to see what's going
>> on and help you.
>>
>> Cheers
>> Joris
>>
>> On Wed, Jun 2, 2010 at 2:21 AM, Changbin Du <changb...@gmail.com> wrote:
>>
>>> HI, Dear R community,
>>>
>>> I am using the following codes to do the som. I tried to label the notes
>>> by
>>> the majority vote. either through mapping or prediction.
>>> I attached my output, the left one dont have any labels in the note, the
>>> right one has more than one label in each note. I need to have only one
>>> label for each note either by majority vote or prediction.
>>>
>>> Can anyone give some suggestions or advice? Thanks so much!
>>>
>>>
>>>
>>> alex<-read.table("/home/cdu/operon/alex2.txt", , sep="\t", skip=0,
>>> header=T,
>>> fill=T)
>>> alex1<-alex[,c(1:257)]
>>> levels(alex1$Label)
>>>
>>> alex1$outcome<-as.numeric(alex1$Label)
>>> alex1$outcome[1:20]
>>>
>>>
>>> #self-organizing maps(unsupervised learning)
>>> library("kohonen")
>>>
>>>
>>> #SOM, the supervised learning, train the map using outcome as the class
>>> variable.
>>> set.seed(13)
>>> final.xyf<- xyf(data=as.matrix(alex1[,c(1:256)]),
>>> Y=classvec2classmat(alex1$outcome), xweight = 0.99, grid=somgrid(20, 30,
>>> "hexagonal"))
>>>
>>>
>>> outcome.xyf <- predict(final.xyf)$unit.prediction#get prediction
>>> outcome.predict<- as.numeric(classmat2classvec(outcome.xyf)) #change
>>> matrix
>>> to vectors.
>>>
>>> outcome.label<-LETTERS[outcome.predict] #conver the numeric value to
>>> letters.
>>>
>>>
>>> plot(final.xyf, type="property", property=outcome.predict,
>>> labels=outcome.label, palette.name =rainbow, main="Prediction ")
>>>
>>>
>>>
>>> cl <- colors()
>>> bgcols <- cl[2:14]
>>> plot(final.xyf, type="mapping", labels=outcome.label, col="black",
>>> bgcol=bgcols[as.integer(outcome.predict)],
>>> main="Mapping plot")
>>>
>>>
>>>
>>>
>>> --
>>> Sincerely,
>>> Changbin
>>> --
>>>
>>> ______________________________________________
>>> R-help@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.
>>>
>>>
>>
>>
>> --
>> Joris Meys
>> Statistical Consultant
>>
>> Ghent University
>> Faculty of Bioscience Engineering
>> Department of Applied mathematics, biometrics and process control
>>
>> Coupure Links 653
>> B-9000 Gent
>>
>> tel : +32 9 264 59 87
>> joris.m...@ugent.be
>> -------------------------------
>> Disclaimer : http://helpdesk.ugent.be/e-maildisclaimer.php
>>
>
>
>
> --
> Sincerely,
> Changbin
> --
>
>
>
--
Joris Meys
Statistical Consultant
Ghent University
Faculty of Bioscience Engineering
Department of Applied mathematics, biometrics and process control
Coupure Links 653
B-9000 Gent
tel : +32 9 264 59 87
joris.m...@ugent.be
-------------------------------
Disclaimer : http://helpdesk.ugent.be/e-maildisclaimer.php
[[alternative HTML version deleted]]
______________________________________________
R-help@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.
