tests/testthat/test-plotting.R

In bioLeak: Leakage-Safe Modeling and Auditing for Genomic and Clinical Data

test_that("plotting helpers validate ggplot2 availability", {
  skip_if_not_installed("ggplot2")
  df <- make_class_df(10)
  splits <- make_split_plan_quiet(df, outcome = "outcome",
                              mode = "subject_grouped", group = "subject",
                              v = 2, seed = 1)
  custom <- make_custom_learners()
  fit <- fit_resample_quiet(df, outcome = "outcome", splits = splits,
                            learner = "glm", custom_learners = custom,
                            metrics = "accuracy", refit = FALSE, seed = 1)
  audit <- audit_leakage(fit, metric = "auc", B = 3, perm_stratify = FALSE)

  if (requireNamespace("ggplot2", quietly = TRUE)) {
    expect_true(TRUE)
  } else {
    expect_error(plot_perm_distribution(audit), "ggplot2")
    expect_error(plot_fold_balance(fit), "ggplot2")
    expect_error(plot_overlap_checks(fit, column = "subject"), "ggplot2")
    expect_error(plot_time_acf(fit), "ggplot2")
  }
})

test_that("plotting helpers return objects when ggplot2 is installed", {
  skip_if_not_installed("ggplot2")
  df <- make_class_df(12)
  splits <- make_split_plan_quiet(df, outcome = "outcome",
                              mode = "subject_grouped", group = "subject",
                              v = 3, seed = 1)
  custom <- make_custom_learners()
  fit <- fit_resample_quiet(df, outcome = "outcome", splits = splits,
                            learner = "glm", custom_learners = custom,
                            metrics = "auc", refit = FALSE, seed = 1)
  audit <- audit_leakage(fit, metric = "auc", B = 5, perm_stratify = FALSE)

  with_temp_plot_device({
    res1 <- plot_perm_distribution(audit)
    expect_true(all(c("observed", "permuted_mean", "plot") %in% names(res1)))

    res2 <- plot_fold_balance(fit)
    expect_true(all(c("fold_summary", "plot") %in% names(res2)))

    res3 <- plot_overlap_checks(fit, column = "subject")
    expect_true(all(c("overlap_counts", "plot") %in% names(res3)))

    # plot_time_acf requires time_series splits; skip here
    expect_error(plot_time_acf(fit, lag.max = 3), "time")
  })
})

test_that("plot_overlap_checks and plot_time_acf validate inputs", {
  skip_if_not_installed("ggplot2")
  df <- make_class_df(8)
  df <- df[, c("subject", "outcome", "x1", "x2")]
  splits <- make_split_plan_quiet(df, outcome = "outcome",
                              mode = "subject_grouped", group = "subject",
                              v = 2, seed = 1)
  custom <- make_custom_learners()
  fit <- fit_resample_quiet(df, outcome = "outcome", splits = splits,
                            learner = "glm", custom_learners = custom,
                            metrics = "accuracy", refit = FALSE, seed = 1)
  expect_error(plot_overlap_checks(fit, column = "missing"), "Column not available")

  fit@predictions <- list(data.frame(pred = 1, truth = factor(1), id = 1, fold = 1, learner = "glm"))
  expect_error(plot_time_acf(fit, lag.max = 2), "time")
})

test_that("plot_perm_distribution errors without permutation values", {
  skip_if_not_installed("ggplot2")
  df <- make_class_df(10)
  splits <- make_split_plan_quiet(df, outcome = "outcome",
                              mode = "subject_grouped", group = "subject",
                              v = 2, seed = 1)
  custom <- make_custom_learners()
  fit <- fit_resample_quiet(df, outcome = "outcome", splits = splits,
                            learner = "glm", custom_learners = custom,
                            metrics = "auc", refit = FALSE, seed = 1)
  audit <- audit_leakage(fit, metric = "auc", B = 3, return_perm = FALSE,
                         perm_stratify = FALSE)
  expect_error(plot_perm_distribution(audit), "No finite permutation values")
})