R/Holdout.R

In houzeyu2683/PyKit:

##
##  Holdout
Holdout <- function(list, prabability = 0.3, seed = 0){


  ##
  ##  Load package
  require("PyKit")


  ##
  ##  Set random seed
  set.seed(seed)


  ##
  ##  Check list name
  if(is.null(names(list)))
    names(list) <- seq(length(list))

  ##
  ##  Total data
  intFirstIndex       <- 1
  booHaveDimension    <- HaveDimension(list[[intFirstIndex]])
  if(booHaveDimension){
    intTotal <- nrow(list[[1]])
  }else{
    intTotal <- length(list[[1]])
  }


  ##
  ##  Sample size
  intSampleSize  <- floor(intTotal * prabability)
  eleSampleIndex <- sample(seq(intTotal), intSampleSize)


  ##
  ##  return
  listReturn <- list()
  for(i in names(list)){
    booHaveDimension    <- HaveDimension(list[[i]])
    if(booHaveDimension){
      listReturn[['Train']][[i]] <- list[[i]][-eleSampleIndex,]
      listReturn[['Valid']][[i]] <- list[[i]][ eleSampleIndex,]
    }else{
      listReturn[['Train']][[i]] <- list[[i]][-eleSampleIndex]
      listReturn[['Valid']][[i]] <- list[[i]][ eleSampleIndex]
    }
  }
  return(listReturn)
}