old_C_code/stage2_UNFINISHED/src/05_cluster2stepWER.r

   1 ## File : 05_cluster2stepWER.r
   2 ## Description :
   3
   4 rm(list = ls())
   5
   6 setwd("~/ownCloud/projects/2014_EDF-Orsay-Lyon2/codes/")
   7
   8 library(Rwave)       # CWT
   9 library(cluster)     # pam
  10 #library(flexclust)   # kcca
  11 source("aux.r")               # auxiliary clustering functions
  12
  13 #TODO: [plus tard] alternative à sowa (package disparu) : cwt..
  14 source("sowas-superseded.r")  # auxiliary CWT functions
  15
  16 ## 1. Read auxiliar data files ####
  17
  18 identifiants <- read.table("identifs.txt")[ ,1]
  19 dates0       <- read.table("datesall.txt")[, 1]
  20 dates        <- as.character(dates0[grep("2009", dates0)])
  21 rm(dates0)
  22
  23 n     <- length(identifiants)
  24 p <- delta <- length(dates)
  25
  26 synchros09 <- t(as.matrix(read.table("~/tmp/2009_synchros200RC.txt")))
  27 #synchros09 <- t(as.matrix(read.table("~/tmp/2009_synchros200-random.txt")))
  28
  29 nas <- which(is.na(synchros09)[, 1]) # some 1/1/2009 are missing
  30 synchros09[nas, 1] <- rowMeans(synchros09[nas, 2:4]) #valeurs après 1er janvier
  31
  32 #moyenne pondérée pour compléter deux demi-heures manquantes
  33 imput09 <- synchros09[, 4180:4181] %*% matrix(c(2/3, 1/3, 1/3, 2/3), 2)
  34 synchros09 <- cbind(synchros09[, 1:4180], imput09, synchros09[, 4181:17518])
  35
  36 conso <- synchros09[-201, ];  # series must be on rows
  37 n     <- nrow(conso)
  38 delta <- ncol(conso)
  39
  40 rm(synchros09, nas)
  41
  42 ## 2. Compute WER distance matrix ####
  43
  44 ## _.a CWT -- Filtering the lowest freqs (>6m) ####
  45 # nvoice   <- 4
  46 # # noctave4 = 2^13 = 8192 half hours ~ 180 days
  47 # noctave4 <- adjust.noctave(N = delta, dt = 1, s0 = 2,
  48 #                           tw = 0, noctave = 13)
  49 # # 4 here represent 2^5 = 32 half-hours ~ 1 day
  50 # scalevector4  <- 2^(4:(noctave4 * nvoice) / nvoice) * 2
  51 # lscvect4      <- length(scalevector4)
  52 # lscvect <- lscvect4  # i should clean my code: werFam demands a lscvect
  53
  54
  55 #17000 colonnes coeff 1, puis 17000 coeff 2... [non : dans chaque tranche du cube]
  56
  57 #TODO: une fonction qui fait lignes 59 à 91
  58
  59 #cube:
  60 # Xcwt4   <- toCWT(conso, noctave = noctave4, dt = 1,
  61 #                 scalevector = scalevector4,
  62 #                 lt = delta, smooth = FALSE,
  63 #                 nvoice = nvoice)      # observations node with CWT
  64 #
  65 # #matrix:
  66 # ############Xcwt2 <- matrix(0.0, nrow= n, ncol= 2 + delta * lscvect)
  67 # #Xcwt2 <- matrix(NA_complex_, nrow= n, ncol= 2 + length((c(Xcwt4[,,1]))))
  68 #
  69 # #NOTE: delta et lscvect pourraient etre gardés à part (communs)
  70 # for(i in 1:n)
  71 #  Xcwt2[i,] <- c(delta, lscvect, Xcwt4[,,i] / max(Mod(Xcwt4[,,i])) )
  72 #
  73 # #rm(conso, Xcwt4); gc()
  74 #
  75 # ## _.b WER^2 distances  ########
  76 # Xwer_dist    <- matrix(0.0, n, n)
  77 # for(i in 1:(n - 1)){
  78 #  mat1   <- vect2mat(Xcwt2[i,])
  79 #  for(j in (i + 1):n){
  80 #     mat2 <- vect2mat(Xcwt2[j,])
  81 #     num     <- Mod(mat1 * Conj(mat2))
  82 #     WX      <- Mod(mat1 * Conj(mat1))
  83 #     WY      <- Mod(mat2 * Conj(mat2))
  84 #     smsmnum <- smCWT(num, scalevector = scalevector4)
  85 #     smsmWX  <- smCWT(WX,  scalevector = scalevector4)
  86 #     smsmWY  <- smCWT(WY,  scalevector = scalevector4)
  87 #     wer2    <- sum(colSums(smsmnum)^2)  /
  88 #       sum( sum(colSums(smsmWX) * colSums(smsmWY)) )
  89 #     Xwer_dist[i, j] <- sqrt(delta * lscvect * (1 - wer2))
  90 #     Xwer_dist[j, i] <- Xwer_dist[i, j]
  91 #   }
  92 # }
  93 # diag(Xwer_dist) <- numeric(n)
  94 #
  95 # save(Xwer_dist, file = "../res/2009_synchros200WER.Rdata")
  96 # save(Xwer_dist, file = "../res/2009_synchros200-randomWER.Rdata")
  97
  98 load("../res/2009_synchros200WER.Rdata")
  99 #load("../res/2009_synchros200-randomWER.Rdata")
 100
 101 ## 3. Cluster using WER distance matrix ####
 102
 103 #hc    <- hclust(as.dist(Xwer_dist), method = "ward.D")
 104 #plot(hc)
 105 #
 106 # #clust <- cutree(hc, 2)
 107 #
 108 for(K in 2:30){
 109   #K <- 3
 110   #pamfit <- pam(tdata[-201, ci$selectv], k = K)
 111   pamfit <- pam(as.dist(Xwer_dist), k = K, diss = TRUE)
 112
 113   #table(pamfit$clustering)
 114
 115   SC <- matrix(0, ncol = p, nrow = K)
 116
 117   clustfactor <- pamfit$clustering
 118 #  for(k in 1:K){
 119 #    clustk <- which(clustfactor == k)
 120 #  if(length(clustk) > 0) {
 121 #    if(length(clustk) > 1) {
 122 #      SCk <- colSums(synchros09[which(clustfactor == k), ])
 123 #      } else {
 124 #        SCk <- synchros09[which(clustfactor == k), ]
 125 #      }
 126 #      SC[k, ] <- SC[k, ] + SCk
 127 #      rm(SCk)
 128 #  }
 129 #}
 130
 131 #write.table(clustfactor, file = paste0("~/tmp/clustfactorRC", K, ".txt"))
 132 #write.table(clustfactor, file = "~/tmp/clustfactor3.txt")
 133 #write.table(clustfactor, file = paste0("~/tmp/clustfactorWER", K, ".txt"))
 134 write.table(clustfactor, file = paste0("~/tmp/clustfactor-randomWER", K, ".txt"))
 135 }
 136 #
 137 # # Plots
 138 # layout(1)
 139 # matplot(t(SC)[48*10 + 1:(48*30), ],  type = 'l', ylab = '',col = 1:3, lty = 1)
 140 # matplot(t(SC)[48*100 + 1:(48*30), ], type = 'l', ylab = '', col = 1:3, lty = 1)
 141 #
 142 #
 143 #