| Clustions {lingua} | R Documentation |
Clustions are for notion-wise clustering of [0,1]-valued data, possibly with the [-1,1] interval being used as compressed pairs of dichotomic data.
clustions.thresholds(thresh) clustions.maxcycles(maxcyc) clustions.cluster(data.cases, ini.centers, by.rows=FALSE) clustions.cluster2(data.cases, ini.centers, by.rows=FALSE)
thresh |
either none or [0,1] value for the threshold |
maxcyc |
either none or the maximal k-means cycling count |
data.cases |
[0,1]-valued data vectors to aggregate |
ini.centers |
initial centers for the clustering |
by.rows |
having row-wise data properties to cluster |
The clustering proceeds like k-means with the similarity metrics based on Contifiers, especially the product ones. Cases with zero-only values are neglected, greater similar values count for, greater dissimilar values count against similarities.
thresholds and maxcyc give the set option values,
cluster and cluster2 give the clustering results.
Look at the Bioplexity and Enduce webs for more information. http://www.bioplexity.org/analysis/ http://www.tangloid.net/enduce/
Martin Saturka
Bioplexity www.bioplexity.org
See Also lingua, dinorms, contifiers
library(lingua) ## initial settings contifiers.thresholds(c(0.1)) clustions.maxcycles(20) ## some random data matrix data.cases <- matrix(runif(30), nrow=5, ncol=6) ## with some initial centers ini.centers <- data.cases[,1:3] ## and to cluster it clustions.cluster(data.cases, ini.centers) ## some [-1,1]-valued data (viewed as compressed pairs) data.cases2 <- matrix(runif(30, -1, 1), nrow=5, ncol=6) ## some initial centers ini.cens2neg <- -1 * data.cases2[,1:3] ini.cens2neg[0.0 > ini.cens2neg] <- 0.0 ini.cens2pos <- data.cases2[,1:3] ini.cens2pos[0.0 > ini.cens2pos] <- 0.0 ini.centers2 <- cbind(ini.cens2neg, ini.cens2pos) ## and to cluster it clustions.cluster2(data.cases2, ini.centers2) ## columns were used as objects to cluster