R/GM_Loglikelihood.R

In CorReg: Linear Regression Based on Linear Structure Between Variables

# ' Compute the log-likelihood of a gaussian mixture value due to missing values in a regression
# ' @param X is the vector of the regressors
# ' @param Y is the value of the response variable (scalar)
# ' @param B is the vector of regression parameters, including the intercept in first position  (zero if no intercept)
# ' @param sigma is the standard deviation of the residual of the subregression
# ' @param M is the indicatrice vector (submatrix) of the missing values
# ' @param mixmod is result of calcul_BIC_mixmod2.0(X=dataset,nbclustmax=nbclustmax,details=TRUE)
# ' @param log boolean to define if you want the likelihood or the log-likelihood
# ' @param intercept boolean to define if there is an intercept in the model (default=TRUE)
GM_Loglikelihood <- function(Y = Y, X = X, B = B, sigma = sigma, M = NULL, mixmod, log = TRUE, intercept = TRUE) {
  if (is.null(M)) {
    M = 0 * X
    M[is.na(M)] = 1
    X[is.na(X)] = 0
  }
  return(.Call("GM_likelihood", Y, X, B, sigma, M, mixmod$nbclust, mixmod$details, log, intercept, PACKAGE = "CorReg"))
}