Cluster relabeling function (useful for confusion tables with clusters labeled with k-means for example). runs in complexity O(n^2*m^2) with n the length of labels1 and m the number of clusters.

relabel.R

# arguments :
#    labels1 : list of labels corresponding to the wrong clusters
#    labels2 : list of labels corresponding to the right clusters
# return :
#    labels1 relabeled
relabel <- function(labels1, labels2){
  newlabels <- labels1
  levels1 <- levels(as.factor(labels1))
  levels2 <- levels(as.factor(labels2))
  for(i in levels1){
    max_i_eq_j <- c(0,0)
    for(j in levels2){
      n_i_eq_j <- length(which(which(labels1==i) %in% which(labels2==j)))
      if(n_i_eq_j>as.numeric(max_i_eq_j[1])){
        max_i_eq_j[1]<-n_i_eq_j
        max_i_eq_j[2]<-j
      }
    }
    newlabels[which(labels1==i)] <- max_i_eq_j[2]
  }
  return(newlabels)
}

This code is released under the .