#' EMGrank
#'
-#' Run an generalized EM algorithm developped for mixture of Gaussian regression
+#' Run an generalized EM algorithm developped for mixture of Gaussian regression
#' models with variable selection by an extension of the low rank estimator.
#' Reparametrization is done to ensure invariance by homothetic transformation.
#' It returns a collection of models, varying the number of clusters and the rank of the regression mean.
#' @param Y matrix of responses (of size n*m)
#' @param eps real, threshold to say the EM algorithm converges, by default = 1e-4
#' @param rank vector of possible ranks
+#' @param fast boolean to enable or not the C function call
#'
#' @return A list (corresponding to the model collection) defined by (phi,LLF):
-#' phi : regression mean for each cluster
+#' phi : regression mean for each cluster, an array of size p*m*k
#' LLF : log likelihood with respect to the training set
#'
#' @export
-EMGrank <- function(Pi, Rho, mini, maxi, X, Y, eps, rank, fast = TRUE)
+EMGrank <- function(Pi, Rho, mini, maxi, X, Y, eps, rank, fast)
{
if (!fast)
{
}
# Function in C
- n <- nrow(X) #nombre d'echantillons
- p <- ncol(X) #nombre de covariables
- m <- ncol(Y) #taille de Y (multivarie)
- k <- length(Pi) #nombre de composantes dans le melange
- .Call("EMGrank", Pi, Rho, mini, maxi, X, Y, eps, as.integer(rank), phi = double(p * m * k),
- LLF = double(1), n, p, m, k, PACKAGE = "valse")
+ .Call("EMGrank", Pi, Rho, mini, maxi, X, Y, eps, as.integer(rank), PACKAGE = "valse")
}
# helper to always have matrices as arg (TODO: put this elsewhere? improve?) -->
# Yes, we should use by-columns storage everywhere... [later!]
matricize <- function(X)
{
- if (!is.matrix(X))
+ if (!is.matrix(X))
return(t(as.matrix(X)))
- return(X)
+ X
}
# R version - slow but easy to read
for (r in 1:k)
{
Z_indice <- seq_len(n)[Z == r] #indices where Z == r
- if (length(Z_indice) == 0)
+ if (length(Z_indice) == 0)
next
# U,S,V = SVD of (t(Xr)Xr)^{-1} * t(Xr) * Yr
- s <- svd(MASS::ginv(crossprod(matricize(X[Z_indice, ]))) %*%
- crossprod(matricize(X[Z_indice, ]), matricize(Y[Z_indice, ])))
+ s <- svd(MASS::ginv(crossprod(matricize(X[Z_indice, ]))) %*%
+ crossprod(matricize(X[Z_indice, ]), matricize(Y[Z_indice, ])))
S <- s$d
# Set m-rank(r) singular values to zero, and recompose best rank(r) approximation
# of the initial product
- if (rank[r] < length(S))
+ if (rank[r] < length(S))
S[(rank[r] + 1):length(S)] <- 0
phi[, , r] <- s$u %*% diag(S) %*% t(s$v) %*% Rho[, , r]
}
# update distance parameter to check algorithm convergence (delta(phi, Phi))
deltaPhi <- c(deltaPhi, max((abs(phi - Phi))/(1 + abs(phi)))) #TODO: explain?
- if (length(deltaPhi) > deltaPhiBufferSize)
+ if (length(deltaPhi) > deltaPhiBufferSize)
deltaPhi <- deltaPhi[2:length(deltaPhi)]
sumDeltaPhi <- sum(abs(deltaPhi))
Phi <- phi
ite <- ite + 1
}
- return(list(phi = phi, LLF = LLF))
+ list(phi = phi, LLF = LLF)
}