update Emilie
[valse.git] / pkg / R / main.R
... / ...
CommitLineData
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
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 a list with estimators of parameters
27#'
28#' @examples
29#' #TODO: a few examples
30#'
31#' @export
32runValse <- function(X, Y, procedure = "LassoMLE", selecMod = "DDSE", gamma = 1, mini = 10,
33 maxi = 50, eps = 1e-04, kmin = 2, kmax = 3, rank.min = 1, rank.max = 5, ncores_outer = 1,
34 ncores_inner = 1, thresh = 1e-08, grid_lambda = numeric(0), size_coll_mod = 10,
35 fast = TRUE, verbose = FALSE, plot = TRUE)
36{
37 n <- nrow(X)
38 p <- ncol(X)
39 m <- ncol(Y)
40
41 if (verbose)
42 print("main loop: over all k and all lambda")
43
44 if (ncores_outer > 1) {
45 cl <- parallel::makeCluster(ncores_outer, outfile = "")
46 parallel::clusterExport(cl = cl, envir = environment(), varlist = c("X",
47 "Y", "procedure", "selecMod", "gamma", "mini", "maxi", "eps", "kmin",
48 "kmax", "rank.min", "rank.max", "ncores_outer", "ncores_inner", "thresh",
49 "size_coll_mod", "verbose", "p", "m"))
50 }
51
52 # Compute models with k components
53 computeModels <- function(k)
54 {
55 if (ncores_outer > 1)
56 require("valse") #nodes start with an empty environment
57
58 if (verbose)
59 print(paste("Parameters initialization for k =", k))
60 # smallEM initializes parameters by k-means and regression model in each
61 # component, doing this 20 times, and keeping the values maximizing the
62 # likelihood after 10 iterations of the EM algorithm.
63 P <- initSmallEM(k, X, Y, fast)
64 if (length(grid_lambda) == 0)
65 {
66 grid_lambda <- computeGridLambda(P$phiInit, P$rhoInit, P$piInit, P$gamInit,
67 X, Y, gamma, mini, maxi, eps, fast)
68 }
69 if (length(grid_lambda) > size_coll_mod)
70 grid_lambda <- grid_lambda[seq(1, length(grid_lambda), length.out = size_coll_mod)]
71
72 if (verbose)
73 print("Compute relevant parameters")
74 # select variables according to each regularization parameter from the grid:
75 # S$selected corresponding to selected variables
76 S <- selectVariables(P$phiInit, P$rhoInit, P$piInit, P$gamInit, mini, maxi,
77 gamma, grid_lambda, X, Y, thresh, eps, ncores_inner, fast)
78
79 if (procedure == "LassoMLE") {
80 if (verbose)
81 print("run the procedure Lasso-MLE")
82 # compute parameter estimations, with the Maximum Likelihood Estimator,
83 # restricted on selected variables.
84 models <- constructionModelesLassoMLE(P$phiInit, P$rhoInit, P$piInit,
85 P$gamInit, mini, maxi, gamma, X, Y, eps, S, ncores_inner, fast, verbose)
86 } else {
87 if (verbose)
88 print("run the procedure Lasso-Rank")
89 # compute parameter estimations, with the Low Rank Estimator, restricted on
90 # selected variables.
91 models <- constructionModelesLassoRank(S, k, mini, maxi, X, Y, eps, rank.min,
92 rank.max, ncores_inner, fast, verbose)
93 }
94 # warning! Some models are NULL after running selectVariables
95 models <- models[sapply(models, function(cell) !is.null(cell))]
96 models
97 }
98
99 # List (index k) of lists (index lambda) of models
100 models_list <-
101 if (ncores_outer > 1) {
102 parLapply(cl, kmin:kmax, computeModels)
103 } else {
104 lapply(kmin:kmax, computeModels)
105 }
106 if (ncores_outer > 1)
107 parallel::stopCluster(cl)
108
109 if (!requireNamespace("capushe", quietly = TRUE))
110 {
111 warning("'capushe' not available: returning all models")
112 return(models_list)
113 }
114
115 # Get summary 'tableauRecap' from models
116 tableauRecap <- do.call(rbind, lapply(seq_along(models_list), function(i)
117 {
118 models <- models_list[[i]]
119 # For a collection of models (same k, several lambda):
120 LLH <- sapply(models, function(model) model$llh[1])
121 k <- length(models[[1]]$pi)
122 sumPen <- sapply(models, function(model) k * (dim(model$rho)[1] + sum(model$phi[,
123 , 1] != 0) + 1) - 1)
124 data.frame(model = paste(i, ".", seq_along(models), sep = ""), pen = sumPen/n,
125 complexity = sumPen, contrast = -LLH)
126 }))
127 tableauRecap <- tableauRecap[which(tableauRecap[, 4] != Inf), ]
128
129 if (verbose == TRUE)
130 print(tableauRecap)
131 modSel <- capushe::capushe(tableauRecap, n)
132 indModSel <- if (selecMod == "DDSE")
133 {
134 as.numeric(modSel@DDSE@model)
135 } else if (selecMod == "Djump")
136 {
137 as.numeric(modSel@Djump@model)
138 } else if (selecMod == "BIC")
139 {
140 modSel@BIC_capushe$model
141 } else if (selecMod == "AIC")
142 {
143 modSel@AIC_capushe$model
144 }
145
146 listMod <- as.integer(unlist(strsplit(as.character(indModSel), "[.]")))
147 modelSel <- models_list[[listMod[1]]][[listMod[2]]]
148 modelSel$tableau <- tableauRecap
149
150 if (plot)
151 print(plot_valse(X, Y, modelSel, n))
152
153 return(modelSel)
154}