R/mhidetify.R

In hidetify: Identify Influential Observations in High Dimension

### Nous avons remplace la fonction asymMIP par hidetify


mhidetify <- function(
                        x, 
                        y, 
                        number_subset, 
                        size_subset, 
                        asymvec, 
                        ep = 0.1, 
                        alpha_swamp, 
                        alpha_mask, 
                        alpha_validate
                        )
{
  df <- length(asymvec)
  xquant <- apply(x,2,quantile,asymvec)
  yquant <- quantile(y,asymvec)
  inv_rob_sdx <- 1/apply(x,2,mad)
  rob_sdy <- mad(y)
  xnrow <- nrow(x)
  est_clean_set <- c(1:xnrow)
  est_clean_set <- ease_swamping(x,y,xquant, yquant, inv_rob_sdx, 
                                 rob_sdy,number_subset,size_subset,
                                 est_clean_set,asymvec,ep <- 0.1,
                                 alpha_swamp)$est_clean_set
  clean_set <- ease_masking(x,y, xquant, yquant, inv_rob_sdx, 
                            rob_sdy, number_subset,size_subset,
                            est_clean_set,asymvec, 
                            alpha_mask)$clean_set

  while (length(clean_set) < xnrow/2)
  {
    est_clean_set <- ease_swamping(x,y,xquant, yquant, inv_rob_sdx, 
                                   rob_sdy,number_subset,size_subset,
                                   est_clean_set,asymvec,ep <- 0.1,
                                   alpha_swamp)$est_clean_set
    clean_set <- ease_masking(x,y, xquant, yquant, inv_rob_sdx, 
                              rob_sdy, number_subset,size_subset,
                              est_clean_set,asymvec, 
                              alpha_mask)$clean_set
  }

  non_inf_set <- clean_set
  inf_set <- setdiff(c(1:xnrow), non_inf_set)
  inf_setfinal <- vhidetify(x, y, xquant, yquant, inv_rob_sdx, rob_sdy,
                            asymvec, inf_set, non_inf_set, 
                            alpha_validate)$inf_setfinal
  dataframe <- data.frame(ind = 1:xnrow, outlier_ind = rep(0,xnrow))
  dataframe$outlier_ind[which(dataframe$ind%in%inf_setfinal)] <- 1
  dataframe

}