R/WhoIs.R

In CorReg: Linear Regression Based on Linear Structure Between Variables

# ' Give the partition implied by a structure
# ' 
# ' @param Z the structure (square matrix)
# ' @param I1,I2,I3 wanted outputs (I1= explicative covariates, I2=redundant covariates, I3=independent covariates)
# ' 
# '@export 
WhoIs <- function(Z = Z, I3 = FALSE, I2 = TRUE, I1 = TRUE){
  res = list()
  if(I2){
    res$I2 = which(colSums(Z) != 0) # Qui est a gauche
  }
  if(I1){
    res$I1 = which(colSums(Z) == 0) #qui n'est pas a gauche
  }
  if(I3){
    res$I3 = which(colSums(Z) == 0 & rowSums(Z) == 0) # qui est totalement isole
  }
  return(res)
}