R/check.data.R

In HansLandsheer/UncertainInterval: Uncertain Interval Methods for Three-Way Cut-Point Determination in Test Results

#' Function to check the dataset of individuals with (1) and without (0) the
#' targeted condition.
#'
#' @param ref The reference standard. A column in a data frame or a vector
#'   indicating the classification by the reference test. The reference standard
#'   must be coded either as 0 (absence of the condition) or 1 (presence of the
#'   condition)
#' @param test The index test or test under evaluation. A column in a dataset or
#'   vector indicating the test results.
#' @param model The model used for estimation. Default = 'kernel'. When model is
#'   kernel or binormal, the test data is checked whether the test has a
#'   sufficient number of different values (>= 20). When model is ordinal, the
#'   test data are checked whether they are ordinal or not. When model is 'none'
#'   the test data are only checked for missing data.
#' @details The first check is whether ref and test have equal length. If not,
#'   checkdata is aborted with an error message. The second check is whether ref
#'   is coded solely with 0 and 1. If not, check.data is aborted and an error
#'   message is shown. The third check is whether ref and test have missing
#'   values. If true, list wise deletion is applied and a warning message is
#'   shown. The fourth check is whether test is continuous or not. If test has
#'   less than 20 different values, a warning message is shown. This test is
#'   omitted when ordinal = TRUE.
#'
#'   This function is called internally from every function that requires data.
#'   An external call is only useful to check warnings and errors.
#'
#' @return Either a valid dataset as data.frame with two variables ref and test
#'   or an error message.
#' @export
#'
#' @examples
#' set.seed(1)
#' ref=c(rep(0,500), rep(1,500))
#' test=c(rnorm(500,0,1), rnorm(500,1,1.2))
#' check.data(ref, test) # model = 'kernel'

check.data <- function(ref, test, model = c('kernel', 'binormal', 'ordinal', 'none')){

  model <- match.arg(model)

  if (length(ref) != length(test)) stop('parameters ref and test have unequal length')

  sel = stats::complete.cases(ref, test) # sum(sel)
  ref=ref[sel]
  test=test[sel]
  if (!all(sel)) warning('The data has missing values. List wise deletion has been applied, but multiple imputation may provide better results.')
  t=table(test)
  # model = 'ordinal'
  # test = factor(c(1,2,3,4,5), order = T)
  if (model != 'none') {
    if (model != 'ordinal') {
      if (length(names(t)) < 20)
        warning(
          'Your test has less than 20 different values. The ordinal data model may be better appplicable.'
        )
    } else if ((is.factor(test) & (!is.ordered(test)))) {
      stop('Your test is not a factor or not ordered.')
    } else if (is.numeric(test)) {
      if (!all(round(test) - test == 0))
        stop('Your test has non-ordinal numeric values.')
    }
  }


   if (length(levels(as.factor(ref))) != 2) stop('More than 2 values detected in ref.')

   if (is.numeric(ref) & any(levels(as.factor(ref)) != c(0, 1))) {
      stop("Your numeric reference standard must be coded as 0 (absence) and 1 (presence). Check categories of ref!")
   }

   return(data.frame(ref=ref, test=test))
}

# ref[1000]=NA
# test=round(test)