R/miss_imp_PMM.R

In ICRISAT-GEMS/SpaTemHTP: Processing and utilization of HTP data

################
# miss_imp_PMM #
################

#' Missing values imputation using PMM
#' 
#' Impute the missing values using predictive mean matching
#' 
#' Missing values are imputed sequentially from the first to the
#' last day.
#'
#' @param data \code{data.frame} of dimension (N_genotype * N_replicate) x N_days
#' containing the measured phenotypic values.
#' 
#' @param plot \code{logical} value. If plot = TRUE, plot an overview of the
#' missing values pattern. Default = TRUE.
#' 
#' @return Return:
#' 
#' \code{data.frame} with missing values imputed.
#'
#' @author Soumyashree Kar, Vincent Garin
#'
#' @references
#' 
#' Rubin, D. B. (1986). Statistical matching using file concatenation with
#' adjusted weights and multiple imputations. Journal of Business & Economic
#' Statistics, 4(1), 87-94.
#'
#' @examples
#'
#'
#' data(SG_PH_data)
#' 
#' data <- outliers_det_boxplot(data = SG_PH_data[, 6:28])
#' 
#' \dontrun{
#' 
#' data <- miss_imp_PMM(data = data)
#' 
#' }
#' 
#' 
#'
#' @export
#'


miss_imp_PMM <- function(data, plot = TRUE){
  
  mice.ip<-data
  col_nm <- colnames(data) # save colnames for later
  colnames(mice.ip) <- paste0('t_', 1:dim(data)[2])
  
  # optional plot
  if(plot){
    
    mice_plot <- aggr(mice.ip, col=c('navyblue','yellow'),
                      numbers=TRUE, sortVars=FALSE,
                      labels=names(data), cex.axis=.7,
                      gap=3, ylab=c("Missing data","Pattern"))
    
  }
  
  # Impute the missing values.
  imputed_Data <- mice(mice.ip, m=5, maxit = 10, method = 'pmm', seed = 500,
                       printFlag = FALSE)
  
  # check imputed values
  OP <- mice ::complete(imputed_Data, 5, include = FALSE)
  
  # put back the column name
  colnames(OP) <- col_nm
  
  return(OP)
}