tests/testthat/test-multivariate.R

In twolodzko/misster: Simple tools for dealing with missing values

context("Multivariate")


test_that("The missing values are replaced for continuous data.", {

  set.seed(123)

  dat <- mtcars
  dat$disp[sample.int(nrow(dat), 5)] <- NA

  expect_false(anyNA(na_lm(dat, disp ~ mpg + drat)$disp))
  expect_false(anyNA(na_glm(dat, disp ~ mpg + drat)$disp))
  expect_false(anyNA(na_rpart(dat, disp ~ mpg + drat)$disp))
  expect_false(anyNA(na_rf(dat, disp ~ mpg + drat)$disp))
  expect_false(anyNA(na_loess(dat, disp ~ mpg + drat)$disp))

})


test_that("The missing values are replaced for categorical data.", {

  set.seed(123)

  dat <- mtcars
  dat$gear[sample.int(nrow(dat), 10)] <- NA
  dat$gear <- as.factor(dat$gear)

  expect_false(anyNA(na_nb(dat, gear ~ mpg + drat)$disp))
  expect_false(anyNA(na_knn(dat, gear ~ mpg + drat)$disp))
  expect_false(anyNA(na_rpart(dat, factor(gear) ~ mpg + drat)$disp))
  expect_false(anyNA(na_rf(dat, factor(gear) ~ mpg + drat)$disp))

})


test_that("Missing values are replaced or passed if it is not possible for continous data, but data does not change the dimmensions.", {

  dat <- mtcars
  n <- nrow(dat)

  dat$disp[1:10] <- NA
  dat$drat[5:15] <- NA

  expect_warning(expect_identical(nrow(na_lm(dat, disp ~ mpg + drat)), n))
  expect_warning(expect_identical(nrow(na_glm(dat, disp ~ mpg + drat)), n))
  expect_warning(expect_identical(nrow(na_rpart(dat, disp ~ mpg + drat)), n))
  # expect_warning(expect_identical(nrow(na_rf(dat, disp ~ mpg + drat)), n))
  expect_warning(expect_identical(nrow(na_loess(dat, disp ~ mpg + drat)), n))

})


test_that("Missing values are replaced or passed if it is not possible for categorical data, but data does not change the dimmensions.", {

  dat <- mtcars
  n <- nrow(dat)

  dat$gear[1:10] <- NA
  dat$drat[5:15] <- NA

  # expect_warning(expect_identical(nrow(na_nb(dat, gear ~ mpg + drat)), n))
  # expect_warning(expect_identical(nrow(na_knn(dat, gear ~ mpg + drat)), n))
  expect_warning(expect_identical(nrow(na_rpart(dat, factor(gear) ~ mpg + drat)), n))
  # expect_warning(expect_identical(nrow(na_rf(dat, factor(gear) ~ mpg + drat)), n))

})