tests/testthat/test-learn_cvpairs.R

In drsimonj/pipelearner: Tidy machine learning pipelines

context("learn_cvpairs")

df <- data.frame(a = letters, b = 1:26)

test_that("default setup", {
  pl1 <- pipelearner(df)
  pl2 <- pipelearner(df) %>% learn_cvpairs()

  expect_equal(dim(pl1$cv_pairs), dim(pl2$cv_pairs))
  expect_equal(nrow(pl1$cv_pairs$train[[1]]), nrow(pl2$cv_pairs$train[[1]]))
})

test_that("uses correct call with .f missing", {
  pl <- pipelearner(df)

  pl_cv <- learn_cvpairs(pl)$cv_pairs
  modelr_cv <- modelr::crossv_mc(df, n = 1L, test = 0.2)

  expect_equal(dim(pl_cv), dim(modelr_cv))
  expect_equal(names(pl_cv), names(modelr_cv))
})

test_that("Uses correct call when calling crossv functions directly", {
  pl <- pipelearner(df)

  pl_cv <- learn_cvpairs(pl, crossv_kfold, k = 3)$cv_pairs
  modelr_cv <- crossv_kfold(df, k = 3)

  expect_equal(dim(pl_cv), dim(modelr_cv))
  expect_equal(names(pl_cv), names(modelr_cv))

  pl_cv <- learn_cvpairs(pl, crossv_mc, n = 4)$cv_pairs
  modelr_cv <- crossv_mc(df, n = 4)

  expect_equal(dim(pl_cv), dim(modelr_cv))
  expect_equal(names(pl_cv), names(modelr_cv))
})

test_that("Coerces pipelearner", {
  lc <- learn_cvpairs(df)
  pl <- pipelearner(df) %>% learn_cvpairs()

  expect_equal(names(lc), names(pl))
  expect_equal(dim(lc$cv_pairs), dim(pl$cv_pairs))
})