tests/testthat/test-tfidf.R

In tidymodels/textrecipes: Extra 'Recipes' for Text Processing

test_data <- tibble(
  text = c(
    "I would not eat them here or there.",
    "I would not eat them anywhere.",
    "I would not eat green eggs and ham.",
    "I do not like them, Sam-I-am."
  )
)

rec <- recipe(~., data = test_data)

test_that("step_tfidf works as intended", {
  rec <- rec |>
    step_tokenize(text) |>
    step_tfidf(text)

  obj <- rec |>
    prep()

  rec_answer <- unname(as.data.frame(bake(obj, new_data = NULL)))

  manual_answer <- unname(
    data.frame(
      am = c(0 / 8, 0 / 6, 0 / 8, 1 / 8) * log(1 + 4 / 1),
      and = c(0 / 8, 0 / 6, 1 / 8, 0 / 8) * log(1 + 4 / 1),
      anywhere = c(0 / 8, 1 / 6, 0 / 8, 0 / 8) * log(1 + 4 / 1),
      do = c(0 / 8, 0 / 6, 0 / 8, 1 / 8) * log(1 + 4 / 1),
      eat = c(1 / 8, 1 / 6, 1 / 8, 0 / 8) * log(1 + 4 / 3),
      eggs = c(0 / 8, 0 / 6, 1 / 8, 0 / 8) * log(1 + 4 / 1),
      green = c(0 / 8, 0 / 6, 1 / 8, 0 / 8) * log(1 + 4 / 1),
      ham = c(0 / 8, 0 / 6, 1 / 8, 0 / 8) * log(1 + 4 / 1),
      here = c(1 / 8, 0 / 6, 0 / 8, 0 / 8) * log(1 + 4 / 1),
      i = c(1 / 8, 1 / 6, 1 / 8, 2 / 8) * log(1 + 4 / 4),
      like = c(0 / 8, 0 / 6, 0 / 8, 1 / 8) * log(1 + 4 / 1),
      not = c(1 / 8, 1 / 6, 1 / 8, 1 / 8) * log(1 + 4 / 4),
      or = c(1 / 8, 0 / 6, 0 / 8, 0 / 8) * log(1 + 4 / 1),
      sam = c(0 / 8, 0 / 6, 0 / 8, 1 / 8) * log(1 + 4 / 1),
      them = c(1 / 8, 1 / 6, 0 / 8, 1 / 8) * log(1 + 4 / 3),
      there = c(1 / 8, 0 / 6, 0 / 8, 0 / 8) * log(1 + 4 / 1),
      would = c(1 / 8, 1 / 6, 1 / 8, 0 / 8) * log(1 + 4 / 3)
    )
  )

  expect_equal(
    as.matrix(rec_answer),
    as.matrix(manual_answer)
  )

  expect_equal(dim(tidy(rec, 2)), c(1, 4))
  expect_equal(dim(tidy(obj, 2)), c(17, 4))
})

test_that("step_tfidf works with vocabulary argument", {
  rec <- rec |>
    step_tokenize(text) |>
    step_tfidf(text, vocabulary = letters)

  obj <- rec |>
    prep()

  expect_length(
    bake(obj, new_data = NULL),
    26
  )
})

test_that("check_name() is used", {
  dat <- test_data
  dat$tfidf_text_i <- dat$text

  rec <- recipe(~., data = dat) |>
    step_tokenize(text) |>
    step_tfidf(text)

  expect_snapshot(
    error = TRUE,
    prep(rec, training = dat)
  )
})

test_that("idf valeus are trained on training data and applied on test data", {
  data <- tibble(text = c("i g", "i i i"))

  rec <- recipe(~text, data) |>
    step_tokenize(text) |>
    step_tfidf(text) |>
    prep()

  expect_equal(
    bake(rec, data |> slice(1)),
    bake(rec, data) |> slice(1)
  )

  expect_equal(
    rec$steps[[2]]$res[[1]],
    c(g = log(1 + 2 / 1), i = log(1 + 2 / 2))
  )
})

test_that("Backwards compatibility with 1592690d36581fc5f4952da3e9b02351b31f1a2e", {
  # Test that recipes trained with version <= 0.5.0 keep previous behavoir and
  # throw warning about the data leakage.

  data <- tibble(text = c("i g", "i i i"))

  rec <- recipe(~text, data) |>
    step_tokenize(text) |>
    step_tfidf(text) |>
    prep()

  rec$steps[[2]]$res <- list(c("g", "i"))

  expect_snapshot(
    expect_equal(
      bake(rec, data) |> slice(1),
      tibble(
        tfidf_text_g = log(1 + 2 / 1) / 2,
        tfidf_text_i = log(1 + 2 / 2) / 2
      )
    )
  )

  expect_snapshot(
    expect_equal(
      bake(rec, data |> slice(1)),
      tibble(
        tfidf_text_g = log(1 + 2 / 2) / 2,
        tfidf_text_i = log(1 + 2 / 2) / 2
      )
    )
  )
})

test_that("bad args", {
  expect_snapshot(
    error = TRUE,
    recipe(~., data = mtcars) |>
      step_tfidf(vocabulary = 1:10) |>
      prep()
  )
  expect_snapshot(
    error = TRUE,
    recipe(~., data = mtcars) |>
      step_tfidf(smooth_idf = "yes") |>
      prep()
  )
  expect_snapshot(
    error = TRUE,
    recipe(~., data = mtcars) |>
      step_tfidf(norm = "yes") |>
      prep()
  )
  expect_snapshot(
    error = TRUE,
    recipe(~., data = mtcars) |>
      step_tfidf(sublinear_tf = "yes") |>
      prep()
  )
  expect_snapshot(
    error = TRUE,
    recipe(~., data = mtcars) |>
      step_tfidf(prefix = NULL) |>
      prep()
  )
})

test_that("sparse = 'yes' works", {
  rec <- recipe(~., data = test_data)

  dense <- rec |>
    step_tokenize(text) |>
    step_tfidf(text, norm = "l1", sparse = "no") |>
    prep() |>
    bake(NULL)
  sparse <- rec |>
    step_tokenize(text) |>
    step_tfidf(text, norm = "l1", sparse = "yes") |>
    prep() |>
    bake(NULL)

  expect_identical(dense, sparse)

  expect_false(any(vapply(dense, sparsevctrs::is_sparse_double, logical(1))))
  expect_true(all(vapply(sparse, sparsevctrs::is_sparse_double, logical(1))))

  dense <- rec |>
    step_tokenize(text) |>
    step_tfidf(text, norm = "l2", sparse = "no") |>
    prep() |>
    bake(NULL)
  sparse <- rec |>
    step_tokenize(text) |>
    step_tfidf(text, norm = "l2", sparse = "yes") |>
    prep() |>
    bake(NULL)

  expect_identical(dense, sparse)

  expect_false(any(vapply(dense, sparsevctrs::is_sparse_double, logical(1))))
  expect_true(all(vapply(sparse, sparsevctrs::is_sparse_double, logical(1))))

  dense <- rec |>
    step_tokenize(text) |>
    step_tfidf(text, norm = "none", sparse = "no") |>
    prep() |>
    bake(NULL)
  sparse <- rec |>
    step_tokenize(text) |>
    step_tfidf(text, norm = "none", sparse = "yes") |>
    prep() |>
    bake(NULL)

  expect_identical(dense, sparse)

  expect_false(any(vapply(dense, sparsevctrs::is_sparse_double, logical(1))))
  expect_true(all(vapply(sparse, sparsevctrs::is_sparse_double, logical(1))))

  dense <- rec |>
    step_tokenize(text) |>
    step_tfidf(text, sublinear_tf = TRUE, sparse = "no") |>
    prep() |>
    bake(NULL)
  sparse <- rec |>
    step_tokenize(text) |>
    step_tfidf(text, sublinear_tf = TRUE, sparse = "yes") |>
    prep() |>
    bake(NULL)

  expect_identical(dense, sparse)

  expect_false(any(vapply(dense, sparsevctrs::is_sparse_double, logical(1))))
  expect_true(all(vapply(sparse, sparsevctrs::is_sparse_double, logical(1))))
})

test_that("sparse argument is backwards compatible", {
  rec <- recipe(~., data = test_data) |>
    step_tokenize(text) |>
    step_tfidf(text, sparse = "no") |>
    prep()

  exp <- bake(rec, test_data)

  # Simulate old recipe
  rec$steps[[1]]$sparse <- NULL

  expect_identical(
    bake(rec, test_data),
    exp
  )
})

test_that(".recipes_toggle_sparse_args works", {
  rec <- recipe(~., data = test_data) |>
    step_tokenize(text) |>
    step_tfidf(text, sparse = "auto")

  exp <- rec |> prep() |> bake(NULL) |> sparsevctrs::sparsity()

  expect_true(.recipes_estimate_sparsity(rec) >= exp)
})

# Infrastructure ---------------------------------------------------------------

test_that("bake method errors when needed non-standard role columns are missing", {
  tokenized_test_data <- recipe(~text, data = test_data) |>
    step_tokenize(text) |>
    prep() |>
    bake(new_data = NULL)

  rec <- recipe(tokenized_test_data) |>
    update_role(text, new_role = "predictor") |>
    step_tfidf(text) |>
    update_role(text, new_role = "potato") |>
    update_role_requirements(role = "potato", bake = FALSE)

  trained <- prep(rec, training = tokenized_test_data, verbose = FALSE)

  expect_snapshot(
    error = TRUE,
    bake(trained, new_data = tokenized_test_data[, -1])
  )
})

test_that("empty printing", {
  rec <- recipe(mpg ~ ., mtcars)
  rec <- step_tfidf(rec)

  expect_snapshot(rec)

  rec <- prep(rec, mtcars)

  expect_snapshot(rec)
})

test_that("empty selection prep/bake is a no-op", {
  rec1 <- recipe(mpg ~ ., mtcars)
  rec2 <- step_tfidf(rec1)

  rec1 <- prep(rec1, mtcars)
  rec2 <- prep(rec2, mtcars)

  baked1 <- bake(rec1, mtcars)
  baked2 <- bake(rec2, mtcars)

  expect_identical(baked1, baked1)
})

test_that("empty selection tidy method works", {
  rec <- recipe(mpg ~ ., mtcars)
  rec <- step_tfidf(rec)

  expect <- tibble(
    terms = character(),
    token = character(),
    weight = double(),
    id = character()
  )

  expect_identical(tidy(rec, number = 1), expect)

  rec <- prep(rec, mtcars)

  expect_identical(tidy(rec, number = 1), expect)
})

test_that("keep_original_cols works", {
  new_names <- c(
    "tfidf_text_am",
    "tfidf_text_and",
    "tfidf_text_anywhere",
    "tfidf_text_do",
    "tfidf_text_eat",
    "tfidf_text_eggs",
    "tfidf_text_green",
    "tfidf_text_ham",
    "tfidf_text_here",
    "tfidf_text_i",
    "tfidf_text_like",
    "tfidf_text_not",
    "tfidf_text_or",
    "tfidf_text_sam",
    "tfidf_text_them",
    "tfidf_text_there",
    "tfidf_text_would"
  )

  rec <- recipe(~text, data = test_data) |>
    step_tokenize(text) |>
    step_tfidf(text, keep_original_cols = FALSE)

  rec <- prep(rec)
  res <- bake(rec, new_data = NULL)

  expect_equal(
    colnames(res),
    new_names
  )

  rec <- recipe(~text, data = test_data) |>
    step_tokenize(text) |>
    step_tfidf(text, keep_original_cols = TRUE)

  rec <- prep(rec)
  res <- bake(rec, new_data = NULL)

  expect_equal(
    colnames(res),
    c("text", new_names)
  )
})

test_that("keep_original_cols - can prep recipes with it missing", {
  rec <- recipe(~text, data = test_data) |>
    step_tokenize(text) |>
    step_tfidf(text)

  rec$steps[[2]]$keep_original_cols <- NULL

  expect_snapshot(
    rec <- prep(rec)
  )

  expect_no_error(
    bake(rec, new_data = test_data)
  )
})

test_that("printing", {
  rec <- rec |>
    step_tokenize(text) |>
    step_tfidf(text)

  expect_snapshot(print(rec))
  expect_snapshot(prep(rec))
})