pkg/R/valse.R

   1 #' Main function
   2 #'
   3 #' @param X matrix of covariates (of size n*p)
   4 #' @param Y matrix of responses (of size n*m)
   5 #' @param procedure among 'LassoMLE' or 'LassoRank'
   6 #' @param selecMod method to select a model among 'SlopeHeuristic', 'BIC', 'AIC'
   7 #' @param gamma integer for the power in the penaly, by default = 1
   8 #' @param mini integer, minimum number of iterations in the EM algorithm, by default = 10
   9 #' @param maxi integer, maximum number of iterations in the EM algorithm, by default = 100
  10 #' @param eps real, threshold to say the EM algorithm converges, by default = 1e-4
  11 #' @param kmin integer, minimum number of clusters, by default = 2
  12 #' @param kmax integer, maximum number of clusters, by default = 10
  13 #' @param rang.min integer, minimum rank in the low rank procedure, by default = 1
  14 #' @param rang.max integer, maximum rank in the
  15 #' @return a list with estimators of parameters
  16 #' @export
  17 #-----------------------------------------------------------------------
  18 valse = function(X,Y,procedure = 'LassoMLE',selecMod = 'BIC',gamma = 1,mini = 10,
  19                  maxi = 100,eps = 1e-4,kmin = 2,kmax = 5,
  20                  rang.min = 1,rang.max = 10) {
  21   ##################################
  22   #core workflow: compute all models
  23   ##################################
  24
  25   p = dim(phiInit)[1]
  26   m = dim(phiInit)[2]
  27   n = dim(X)[1]
  28
  29   tableauRecap = array(, dim=c(1000,4))
  30   cpt = 0
  31   print("main loop: over all k and all lambda")
  32
  33 for (k in kmin:kmax){
  34     print(k)
  35     print("Parameters initialization")
  36     #smallEM initializes parameters by k-means and regression model in each component,
  37     #doing this 20 times, and keeping the values maximizing the likelihood after 10
  38     #iterations of the EM algorithm.
  39     init = initSmallEM(k, X, Y)
  40     phiInit <<- init$phiInit
  41     rhoInit <<- init$rhoInit
  42     piInit  <<- init$piInit
  43     gamInit <<- init$gamInit
  44     source('~/valse/pkg/R/gridLambda.R')
  45     gridLambda <<- gridLambda(phiInit, rhoInit, piInit, gamInit, X, Y, gamma, mini, maxi, eps)
  46
  47     print("Compute relevant parameters")
  48     #select variables according to each regularization parameter
  49     #from the grid: A1 corresponding to selected variables, and
  50     #A2 corresponding to unselected variables.
  51
  52     params = selectiontotale(phiInit,rhoInit,piInit,gamInit,mini,maxi,gamma,gridLambda[seq(1,length(gridLambda), by=3)],X,Y,1e-8,eps)
  53     params2 = selectVariables(phiInit,rhoInit,piInit,gamInit,mini,maxi,gamma,gridLambda[seq(1,length(gridLambda), by=3)],X,Y,1e-8,eps)
  54     ## etrange : params et params 2 sont différents ...
  55
  56     selected <<- params$selected
  57     Rho <<- params$Rho
  58     Pi <<- params$Pi
  59
  60     if (procedure == 'LassoMLE') {
  61       print('run the procedure Lasso-MLE')
  62       #compute parameter estimations, with the Maximum Likelihood
  63       #Estimator, restricted on selected variables.
  64       model[[k]] = constructionModelesLassoMLE(phiInit, rhoInit,piInit,gamInit,mini,maxi,gamma,X,Y,thresh,eps,selected)
  65       LLH = unlist(model[[k]]$llh)[seq(1,2*length(model[[k]]),2)]
  66       D = unlist(model[[k]]$llh)[seq(1,2*length(model[[k]]),2)+1]
  67     } else {
  68       print('run the procedure Lasso-Rank')
  69       #compute parameter estimations, with the Low Rank
  70       #Estimator, restricted on selected variables.
  71       model = constructionModelesLassoRank(Pi, Rho, mini, maxi, X, Y, eps,
  72                                            A1, rank.min, rank.max)
  73
  74       ################################################
  75       ### Regarder la SUITE
  76       phi = runProcedure2()$phi
  77       Phi2 = Phi
  78       if (dim(Phi2)[1] == 0)
  79       {
  80         Phi[, , 1:k,] <<- phi
  81       } else
  82       {
  83         Phi <<- array(0, dim = c(p, m, kmax, dim(Phi2)[4] + dim(phi)[4]))
  84         Phi[, , 1:(dim(Phi2)[3]), 1:(dim(Phi2)[4])] <<- Phi2
  85         Phi[, , 1:k,-(1:(dim(Phi2)[4]))] <<- phi
  86       }
  87     }
  88     tableauRecap[(cpt+1):(cpt+length(model[[k]])), ] = matrix(c(LLH, D, rep(k, length(model[[k]])), 1:length(model[[k]])), ncol = 4)
  89     cpt = cpt+length(model[[k]])
  90   }
  91   print('Model selection')
  92
  93   tableauRecap = array(dim = c())
  94   if (selecMod == 'SlopeHeuristic') {
  95
  96   } else if (selecMod == 'BIC') {
  97     BIC = -2*tableauRecap[,1]+log(n)*tableauRecap[,2]
  98     indMinBIC = which.min(BIC)
  99     return(model[[tableauRecap[indMinBIC,3]]][[tableauRecap[indMinBIC,4]]])
 100   } else if (selecMod == 'AIC') {
 101
 102   }
 103 }