pkg/R/main.R

   1 #' runValse
   2 #'
   3 #' Main function
   4 #'
   5 #' @param X matrix of covariates (of size n*p)
   6 #' @param Y matrix of responses (of size n*m)
   7 #' @param procedure among 'LassoMLE' or 'LassoRank'
   8 #' @param selecMod method to select a model among 'DDSE', 'DJump', 'BIC' or 'AIC'
   9 #' @param gamma integer for the power in the penaly, by default = 1
  10 #' @param mini integer, minimum number of iterations in the EM algorithm, by default = 10
  11 #' @param maxi integer, maximum number of iterations in the EM algorithm, by default = 100
  12 #' @param eps real, threshold to say the EM algorithm converges, by default = 1e-4
  13 #' @param kmin integer, minimum number of clusters, by default = 2
  14 #' @param kmax integer, maximum number of clusters, by default = 10
  15 #' @param rank.min integer, minimum rank in the low rank procedure, by default = 1
  16 #' @param rank.max integer, maximum rank in the low rank procedure, by default = 5
  17 #' @param ncores_outer Number of cores for the outer loop on k
  18 #' @param ncores_inner Number of cores for the inner loop on lambda
  19 #' @param thresh real, threshold to say a variable is relevant, by default = 1e-8
  20 #' @param grid_lambda, a vector with regularization parameters if known, by default numeric(0)
  21 #' @param size_coll_mod (Maximum) size of a collection of models, by default 50
  22 #' @param fast TRUE to use compiled C code, FALSE for R code only
  23 #' @param verbose TRUE to show some execution traces
  24 #' @param plot TRUE to plot the selected models after run
  25 #'
  26 #' @return
  27 #' The selected model (except if the collection of models
  28 #' has less than 11 models, the function returns the collection as it can not select one using Capushe)
  29 #'
  30 #' @examples
  31 #' n = 50; m = 10; p = 5
  32 #' beta = array(0, dim=c(p,m,2))
  33 #' beta[,,1] = 1
  34 #' beta[,,2] = 2
  35 #' data = generateXY(n, c(0.4,0.6), rep(0,p), beta, diag(0.5, p), diag(0.5, m))
  36 #' X = data$X
  37 #' Y = data$Y
  38 #' res = runValse(X, Y, kmax = 5, plot=FALSE)
  39 #' X <- matrix(runif(100), nrow=50)
  40 #' Y <- matrix(runif(100), nrow=50)
  41 #' res = runValse(X, Y, plot=FALSE)
  42 #'
  43 #' @export
  44 runValse <- function(X, Y, procedure = "LassoMLE", selecMod = "DDSE", gamma = 1, mini = 10,
  45   maxi = 50, eps = 1e-04, kmin = 2, kmax = 3, rank.min = 1, rank.max = 5, ncores_outer = 1,
  46   ncores_inner = 1, thresh = 1e-08, grid_lambda = numeric(0), size_coll_mod = 50,
  47   fast = TRUE, verbose = FALSE, plot = TRUE)
  48 {
  49   n <- nrow(X)
  50   p <- ncol(X)
  51   m <- ncol(Y)
  52
  53   if (verbose) print("main loop: over all k and all lambda")
  54
  55   if (ncores_outer > 1) {
  56     cl <- parallel::makeCluster(ncores_outer, outfile = "")
  57     parallel::clusterExport(cl = cl, envir = environment(), varlist = c("X",
  58       "Y", "procedure", "selecMod", "gamma", "mini", "maxi", "eps", "kmin",
  59       "kmax", "rank.min", "rank.max", "ncores_outer", "ncores_inner", "thresh",
  60       "size_coll_mod", "verbose", "p", "m"))
  61   }
  62
  63   # Compute models with k components
  64   computeModels <- function(k) {
  65     if (ncores_outer > 1)
  66       require("valse") #nodes start with an empty environment
  67
  68     if (verbose)
  69       print(paste("Parameters initialization for k =", k))
  70     # smallEM initializes parameters by k-means and regression model in each
  71     # component, doing this 20 times, and keeping the values maximizing the
  72     # likelihood after 10 iterations of the EM algorithm.
  73     P <- initSmallEM(k, X, Y, fast)
  74     if (length(grid_lambda) == 0) {
  75       grid_lambda <- computeGridLambda(P$phiInit, P$rhoInit, P$piInit, P$gamInit,
  76                                        X, Y, gamma, mini, maxi, eps, fast)
  77     }
  78     if (length(grid_lambda) > size_coll_mod)
  79       grid_lambda <- grid_lambda[seq(1, length(grid_lambda), length.out = size_coll_mod)]
  80
  81     if (verbose)
  82       print("Compute relevant parameters")
  83     # select variables according to each regularization parameter from the grid:
  84     # S$selected corresponding to selected variables
  85     S <- selectVariables(P$phiInit, P$rhoInit, P$piInit, P$gamInit, mini, maxi,
  86       gamma, grid_lambda, X, Y, thresh, eps, ncores_inner, fast)
  87
  88     if (procedure == "LassoMLE") {
  89       if (verbose)
  90         print("run the procedure Lasso-MLE")
  91       # compute parameter estimations, with the Maximum Likelihood Estimator,
  92       # restricted on selected variables.
  93       models <- constructionModelesLassoMLE(P$phiInit, P$rhoInit, P$piInit,
  94         P$gamInit, mini, maxi, gamma, X, Y, eps, S, ncores_inner, fast, verbose)
  95     } else {
  96       if (verbose)
  97         print("run the procedure Lasso-Rank")
  98       # compute parameter estimations, with the Low Rank Estimator, restricted on
  99       # selected variables.
 100       models <- constructionModelesLassoRank(S, k, mini, maxi, X, Y, eps, rank.min,
 101         rank.max, ncores_inner, fast, verbose)
 102     }
 103     # warning! Some models are NULL after running selectVariables
 104     models <- models[sapply(models, function(cell) !is.null(cell))]
 105     models
 106   }
 107
 108   # List (index k) of lists (index lambda) of models
 109   models_list <-
 110     if (ncores_outer > 1) {
 111       parallel::parLapply(cl, kmin:kmax, computeModels)
 112     } else {
 113       lapply(kmin:kmax, computeModels)
 114     }
 115   if (ncores_outer > 1) parallel::stopCluster(cl)
 116
 117   if (!requireNamespace("capushe", quietly = TRUE)) {
 118     warning("'capushe' not available: returning all models")
 119     return(models_list)
 120   }
 121
 122   # Get summary 'tableauRecap' from models
 123   tableauRecap <- do.call(rbind, lapply(seq_along(models_list), function(i) {
 124     models <- models_list[[i]]
 125     # For a collection of models (same k, several lambda):
 126     LLH <- sapply(models, function(model) model$llh[1])
 127     k <- length(models[[1]]$pi)
 128     sumPen <- sapply(models, function(model) k * (dim(model$rho)[1] + sum(model$phi[,,1] != 0) + 1) - 1)
 129     data.frame(model = paste(i, ".", seq_along(models), sep = ""), pen = sumPen/n, complexity = sumPen, contrast = -LLH)
 130   }))
 131   tableauRecap <- tableauRecap[which(tableauRecap[, 4] != Inf), ]
 132   if (verbose) print(tableauRecap)
 133
 134   if (nrow(tableauRecap) > 10) {
 135     modSel <- capushe::capushe(tableauRecap, n)
 136     indModSel <- if (selecMod == "DDSE") {
 137       as.numeric(modSel@DDSE@model)
 138     } else if (selecMod == "Djump") {
 139       as.numeric(modSel@Djump@model)
 140     } else if (selecMod == "BIC") {
 141       modSel@BIC_capushe$model
 142     } else if (selecMod == "AIC") {
 143       modSel@AIC_capushe$model
 144     }
 145     listMod <- as.integer(unlist(strsplit(as.character(indModSel), "[.]")))
 146     modelSel <- models_list[[listMod[1]]][[listMod[2]]]
 147     modelSel$models <- tableauRecap
 148
 149     if (plot) plot_valse(X, Y, modelSel)
 150     return(modelSel)
 151   }
 152   tableauRecap
 153 }