R/XSim2008.R

In daosang/PDI2015: Bi-clustering Biological Dataset using Co-similarity

#' Bi-clustering using Co-Similarity (version in 2008)
#'
#' @import lpSolve
#' @description  The χ-Sim algorithm is a co-similarity based approach which builds on the idea of simultaneously generating the similarity matrices SR (between documents) and SC (between words), each of them built on the basis of the other. This method is to implement the algorithm X-SIM in 2008 of Gilles Bission et Feward Sye Hussain. See in the Reference.
#' @param x matrix or dataframe of predictors, of dimension n*p; each row is an observation vector.
#' @param y response variable (1 or 2)
#' @param itr number of iterations.
#' @export
#' @return Return a list of objects.
#' @references
#' Bisson G, Hussain F. \emph{X-sim - A new similarity measure for the co-clustering task}. 7th IEEE International Conference on Machine Learning and Applications (ICMLA), 11-13th Dec. 2008, San Diego, United States.
#' @examples
#' library(PDI2015)
#' # Test with Colon Cancer dataset
#' res = XSim2008(ColonCancer[, -1], ColonCancer[,1], itr = 4)
#'
#' # Test News Groups dataset
#' res1 = XSim2008(NG20SMI$M2[[1]],  NG20SMI$M2[["class"]],  itr = 4)
#' res2 = XSim2008(NG20SMI$M2[[2]],  NG20SMI$M2[["class"]],  itr = 4)
#' res3 = XSim2008(NG20SMI$M5[[1]],  NG20SMI$M5[["class"]],  itr = 4)
#' res4 = XSim2008(NG20SMI$M10[[1]], NG20SMI$M10[["class"]], itr = 4)
#' res5 = XSim2008(NG20SMI$NG1[[1]], NG20SMI$NG1[["class"]], itr = 4)
#' res6 = XSim2008(NG20SMI$NG2[[1]], NG20SMI$NG2[["class"]], itr = 4)
#' res7 = XSim2008(NG20SMI$NG3[[1]], NG20SMI$NG3[["class"]], itr = 4)
#' # Exemple2: Lung Cancer (Gordon et al.2002)
#' tmp = gordon$x
#'idx = c()
#'dem = 1
#'for (i in 1:dim(tmp)[2]) {
#'  if (  (abs(max(tmp[,i])/min(tmp[,i])) < 5) || (abs(max(tmp[,i]) - min(tmp[,i])) < 600)) {
#'    idx[dem] = i
#'    dem = dem + 1
#'  }
#'}
#'
#'tmp2 = tmp[,-idx]
#'dim(tmp2)
#'
#'classs = as.numeric(gordon$y)
#'
#'# Test 4 algo:
#'# 1 .XSIM 2008
#'tt1 = c()
#'for (i in 1:4){
#'  tmp = XSim2008(tmp2, classs, itr = i)
#'  tt1[i] = tmp$accuracy
#'}
#'
#'# 4. XSIM.mod of Hussain
#'tt4 = c()
#'for (i in 1:4) {
#'  tmp = XSim.mod(tmp2, classs, itr = i)
#'  tt4[i] = tmp$accuracy
#'}
#'
#'# Comparaison
#'max(tt1)
#'max(tt4)
#' # Exemple 3: Colon Cancer (Alon et al.1999):
#'tmp = projectPDI2015::ColonCancer[,-1]
#'idx = c()
#'dem = 1
#'for (i in 1:dim(tmp)[2]) {
#'  if (  (abs(max(tmp[,i])/min(tmp[,i])) < 15) || (abs(max(tmp[,i]) - min(tmp[,i])) < 500)) {
#'   idx[dem] = i
#'   dem = dem + 1
#' }
#'}
#'
#'tmp2 = tmp[,-idx]
#'dim(tmp2)
#'
#'
#'# Test 4 algo:
#'# 1 .XSIM 2008
#'tt1 = c()
#'for (i in 1:4){
#' tmp = XSim2008(tmp2, ColonCancer[,1], itr = i)
#' tt1[i] = tmp$accuracy
#'}
#'
#'# 2. XSIM 2010
#'tt2 = c()
#'for (i in 1:4){
#' tmp = XSim2010(tmp2, ColonCancer[,1], itr = i)
#' tt2[i] = tmp$accuracy
#'}
#'
#'# 3. XSIM 2015
#'tt3 = c()
#'for (i in 1:4){
#' tmp = XSim2015(tmp2, ColonCancer[,1], itr = i)
#'  tt3[i] = tmp$accuracy
#'}
#'
#'# 4. XSIM.mod of Hussain
#'tt4 = c()
#'for (i in 1:4) {
#'  tmp = XSim.mod(tmp2, ColonCancer[,1], itr = i)
#'  tt4[i] = tmp$accuracy
#'}
#'
#'# Comparaison
#'max(tt1)
#'# max(tt2)
#'max(tt3)
#'max(tt4)
#'
XSim2008 <- function(x, y, itr = 4) {
  temps <- proc.time()
  this.call = match.call()
  
  x = as.data.frame(x)
  r_clusts = max(y) # number of row (doc) clusters
  dimM_1 = dim(x)[1]
  dimM_2 = dim(x)[2]
  
  # Normalisation M par ligne. Ensuite, sauvegarder dans MR
  rS = as.matrix(rowSums(abs(x))) # sum by ligne
  rS[which(rS == 0)] <- 1       # to avoid divide by zero
  MR = as.matrix(x/rS)
  
  # Normalisation M par colonne. Sauvagarder dans MC
  cS = t(as.matrix(colSums(abs(x)))) # sum by colum
  cS[which(cS == 0)] <- 1        # to avoid divide by zero
  tMC = matrix(1, dimM_1, dimM_2)
  for (i in 1:dimM_2) {
    tMC[,i] = cS[,i]
  }
  MC = as.matrix(x/tMC)
  
  # initialize SR and SC
  SR = diag(dimM_1)
  SC = diag(dimM_2)
  
  # Iterate between SR and SC
  for (ii in 1:itr) {
    print(paste("Traitement en cours l'iteration No: ", toString(ii)));
    SR = MR %*% SC %*% t(MR)
    for (j in 1:dimM_1) {
      SR[j,j] = 1
    }
    
    SC = t(MC) %*% SR %*% MC
    for (k in 1:dimM_2) {
      SC[k,k] = 1
    }
  }
  
  #Calculate the clusters, par row only.
  if (r_clusts == 0) {
    print("You cannot create 0 row clusters. please specify a value for r_clusts (>0)");
  }
  
  # Hierarchical cluster analysis on a set of dissimilarities and methods for analyzing it.
  # http://stat.ethz.ch/R-manual/R-patched/library/stats/html/hclust.html
  disSR = as.matrix(1 - data.frame(SR))
  
  # Hierarchical aggomerative clustering + Ward's method
  XSIM_v2008 = hclust(dist(disSR),method = "ward.D2")
  groups     = cutree(XSIM_v2008, k = r_clusts) # cut tree into "r_clusts" clusters
  
  # Confusion Matrix
  matrix_confusionT = table(data.matrix(y),data.matrix(groups))
  matrix_confusion  = matrix(data.frame(matrix_confusionT)[,3],ncol = r_clusts)
  matrix_confusion  = t(matrix_confusion[nrow(matrix_confusion):1,])
  
  maximum  = lp.assign(matrix_confusion,direction = "max")$objval
  accuracy =  maximum/(length(t(y))[1])
  print(accuracy)
  print("DONE.")
  
  time = round(((proc.time() - temps)[3][[1]]), 4)
  print(paste("Time:",time, "(s)"))
  return(list(call = this.call, accuracy = accuracy, matrix.confusion = matrix_confusion, time = time, SR = SR, SC = SC))
}
# ------- FIN --------------