R/computeMemberIndices.R

In idendr0: Interactive Dendrograms

computeMemberIndices<-function
### Compute indices of observations forming given cluster.
##keyword<<internal
(
    h, ##<< HCA result, usually 'hclust' return value
    clusterIdx ##<< cluster index
) {
    clusterCount<-nrow(h$merge)
    indices<-rep(NA,clusterCount+1)
    indicesCount<-0
    toTraverse<-c(clusterIdx,rep(NA,clusterCount-1))
    toTraverseStart<-1
    toTraverseEnd<-2
    repeat {
        i<-toTraverse[toTraverseStart]
        if (is.na(i)) break
        toTraverse[toTraverseStart]<-NA
        toTraverseStart<-toTraverseStart+1
        if (h$merge[i,1]>0) {
            toTraverse[toTraverseEnd]<-h$merge[i,1]
            toTraverseEnd<-toTraverseEnd+1
        } else {
            indicesCount<-indicesCount+1
            indices[indicesCount]<- -h$merge[i,1]
        }
        if (h$merge[i,2]>0) {
            toTraverse[toTraverseEnd]<-h$merge[i,2]
            toTraverseEnd<-toTraverseEnd+1
        } else {
            indicesCount<-indicesCount+1
            indices[indicesCount]<- -h$merge[i,2]
        }
    }
    return(indices[1:indicesCount])
    ### indices of observations forming cluster 'clusterIdx'
}