Hi trzeszutek, Another approach you might want to look at is the clustergram: http://www.r-statistics.com/2010/06/clustergram-visualization-and-diagnostics-for-cluster-analysis-r-code/
<http://www.r-statistics.com/2010/06/clustergram-visualization-and-diagnostics-for-cluster-analysis-r-code/> Cheers, Tal ----------------Contact Details:------------------------------------------------------- Contact me: tal.gal...@gmail.com | 972-52-7275845 Read me: www.talgalili.com (Hebrew) | www.biostatistics.co.il (Hebrew) | www.r-statistics.com (English) ---------------------------------------------------------------------------------------------- On Wed, Aug 18, 2010 at 6:20 PM, trzeszutek <rzes...@mcmaster.ca> wrote: > > Hello All, > > I'm having some trouble figuring out what the clearest way to plot my > k-means clustering result on an my existing MDS. > > First I performed MDS on my distance matrix (note: I performed k-means on > the MDS coordinates because applying a euclidean distance measure to my raw > data would have been inappropriate) > > canto.MDS<-cmdscale(canto) > > I then figured out what would be my optimum k-value by plotting the within > sums of squares for K1-K15 > > > wss <- (nrow(canto.MDS)-1)*sum(apply(canto.MDS,2,var)) > > wss[2] <- sum(kmeans(canto.MDS,centers=2)$withinss) > > wss[3] <- sum(kmeans(canto.MDS,centers=3)$withinss) > > wss[4] <- sum(kmeans(canto.MDS,centers=4)$withinss) > > wss[5] <- sum(kmeans(canto.MDS,centers=5)$withinss) > > wss[6] <- sum(kmeans(canto.MDS,centers=6)$withinss) > > wss[7] <- sum(kmeans(canto.MDS,centers=7)$withinss) > > wss[8] <- sum(kmeans(canto.MDS,centers=8)$withinss) > > wss[9] <- sum(kmeans(canto.MDS,centers=9)$withinss) > > wss[10] <- sum(kmeans(canto.MDS,centers=10)$withinss) > > wss[11] <- sum(kmeans(canto.MDS,centers=11)$withinss) > > wss[12] <- sum(kmeans(canto.MDS,centers=12)$withinss) > > wss[13] <- sum(kmeans(canto.MDS,centers=13)$withinss) > > wss[14] <- sum(kmeans(canto.MDS,centers=14)$withinss) > > wss[15] <- sum(kmeans(canto.MDS,centers=15)$withinss) > > > plot(1:15, wss, type="b",xlab="Number of Clusters", ylab="Within groups > > sum of squares") > > I found my "elbow" at K=7 so i performed k-means for K=7 and attached the > cluster designations to my MDS data frame: > > > fit<-kmeans(canto.MDS,7) > > canto.kmeans<-data.frame(canto.MDS, fit$cluster) > > Now my problem is that I want to plot the MDS with coloured points > (according to population designation, which I can do) but overlay some sort > of silhouette or oval to make the cluster assignment clear (which will be > different from the population designation). Is there some sort of function > akin to "points" that can do this for me? > > Attached is my plot for within sums of squares and an example of the MDS > plot on which I want to plot cluster assignment. > > Thanks in advance for your assistance, > > Tom > > __________________________________________________________________________________________ > > Tom Rzeszutek > MSc. Candidate > rzes...@mcmaster.ca > > The NeuroArts Lab (neuroarts.org) > McMaster University > Department of Psychology, Neuroscience and Behaviour > 1280 Main St. W > Hamilton, ON, Canada > L8S 4K1 > > > http://r.789695.n4.nabble.com/file/n2330000/withinSS.png > http://r.789695.n4.nabble.com/file/n2330000/MDS.png > > > -- > View this message in context: > http://r.789695.n4.nabble.com/Plotting-K-means-clustering-results-on-an-MDS-tp2330000p2330000.html > Sent from the R help mailing list archive at Nabble.com. > > [[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. > [[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.