tests/testthat/test-class-detection_prevalence.R

In yardstick: Tidy Characterizations of Model Performance

test_that("Calculations are correct - two class", {
  lst <- data_altman()
  pathology <- lst$pathology

  expect_equal(
    detection_prevalence_vec(
      truth = pathology$pathology,
      estimate = pathology$scan
    ),
    (231 + 32) / 344
  )
})

test_that("Calculations are correct - three class", {
  multi_ex <- data_three_by_three()
  micro <- data_three_by_three_micro()

  expect_equal(
    detection_prevalence(multi_ex, estimator = "macro")[[".estimate"]],
    macro_metric(detection_prevalence_binary)
  )
  expect_equal(
    detection_prevalence(multi_ex, estimator = "macro_weighted")[[".estimate"]],
    macro_weighted_metric(detection_prevalence_binary)
  )
  expect_equal(
    detection_prevalence(multi_ex, estimator = "micro")[[".estimate"]],
    with(micro, sum(tp + fp) / sum(n + p))
  )
})

test_that("All interfaces gives the same results", {
  lst <- data_altman()
  pathology <- lst$pathology
  path_tbl <- lst$path_tbl
  path_mat <- unclass(path_tbl)

  exp <- detection_prevalence_vec(pathology$pathology, pathology$scan)

  expect_identical(
    detection_prevalence(path_tbl)[[".estimate"]],
    exp
  )
  expect_identical(
    detection_prevalence(path_mat)[[".estimate"]],
    exp
  )
  expect_identical(
    detection_prevalence(pathology, truth = pathology, estimate = scan)[[
      ".estimate"
    ]],
    exp
  )
})

test_that("Calculations handles NAs", {
  lst <- data_altman()
  pathology <- lst$pathology

  expect_equal(
    detection_prevalence_vec(
      truth = pathology$pathology,
      estimate = pathology$scan_na
    ),
    (230 + 32) / 341
  )
})

test_that("Case weights calculations are correct", {
  df <- data.frame(
    truth = factor(c("x", "x", "y"), levels = c("x", "y")),
    estimate = factor(c("x", "y", "x"), levels = c("x", "y")),
    case_weights = c(1L, 1L, 2L)
  )

  expect_identical(
    detection_prevalence_vec(
      truth = df$truth,
      estimate = df$estimate,
      case_weights = df$case_weights
    ),
    3 / 4
  )
})

test_that("work with class_pred input", {
  skip_if_not_installed("probably")

  cp_truth <- probably::as_class_pred(two_class_example$truth, which = 1)
  cp_estimate <- probably::as_class_pred(two_class_example$predicted, which = 2)

  fct_truth <- two_class_example$truth
  fct_truth[1] <- NA

  fct_estimate <- two_class_example$predicted
  fct_estimate[2] <- NA

  expect_identical(
    detection_prevalence_vec(fct_truth, cp_estimate),
    detection_prevalence_vec(fct_truth, fct_estimate)
  )

  expect_identical(
    detection_prevalence_vec(fct_truth, cp_estimate, na_rm = FALSE),
    NA_real_
  )

  expect_snapshot(
    error = TRUE,
    detection_prevalence_vec(cp_truth, cp_estimate)
  )
})

test_that("works with hardhat case weights", {
  lst <- data_altman()
  df <- lst$pathology
  imp_wgt <- hardhat::importance_weights(seq_len(nrow(df)))
  freq_wgt <- hardhat::frequency_weights(seq_len(nrow(df)))

  expect_no_error(
    detection_prevalence_vec(df$pathology, df$scan, case_weights = imp_wgt)
  )

  expect_no_error(
    detection_prevalence_vec(df$pathology, df$scan, case_weights = freq_wgt)
  )
})

test_that("na_rm argument check", {
  expect_snapshot(
    error = TRUE,
    detection_prevalence(1, 1, na_rm = "yes")
  )
})

test_that("`event_level = 'second'` works", {
  lst <- data_altman()
  pathology <- lst$pathology
  path_tbl <- lst$path_tbl

  expect_equal(
    detection_prevalence_vec(
      pathology$pathology,
      pathology$scan,
      event_level = "second"
    ),
    1 -
      detection_prevalence_vec(
        pathology$pathology,
        pathology$scan,
        event_level = "first"
      )
  )
})

test_that("range values are correct", {
  # You don't hit best case scenario on this metric unless all levels are the
  # first
  range <- metric_range(detection_prevalence)

  df <- tibble::tibble(
    truth = factor(c("A", "A", "B", "B", "B")),
    off = factor(c("B", "B", "A", "A", "A"))
  )

  expect_gte(detection_prevalence_vec(df$truth, df$truth), range[1])
  expect_lte(detection_prevalence_vec(df$truth, df$truth), range[2])
  expect_gte(detection_prevalence_vec(df$truth, df$off), range[1])
  expect_lte(detection_prevalence_vec(df$truth, df$off), range[2])
})