tests/testthat/test-p_direction.R

In BayesERtools: Bayesian Exposure-Response Analysis Tools

if (.if_run_ex_p_dir() && .if_run_ex_covsel()) {
  require(bayestestR)

  d_sim_binom_cov <- d_sim_binom_cov

  id_to_sample <- seq(1, max(d_sim_binom_cov$ID), by = 5)

  df_er_ae_covsel_test <-
    d_sim_binom_cov |>
    dplyr::filter(
      AETYPE == "ae_covsel_test",
      ID %in% id_to_sample
    ) |>
    dplyr::mutate(
      AUCss_1000 = AUCss / 1000, BAGE_10 = BAGE / 10,
      BWT_10 = BWT / 10, BHBA1C_5 = BHBA1C / 5
    )


  # Create  ermod_bin objects for testing -------------------------------------
  set.seed(1234)
  ermod_bin <- dev_ermod_bin(
    data = df_er_ae_covsel_test,
    var_resp = "AEFLAG",
    var_exposure = "AUCss_1000",
    var_cov = "BHBA1C_5",
    verbosity_level = 0,
    # Need large number of samples to calculate p_direction precisely
    chains = 4, iter = 2000
  )

  ermod_bin_cov_sel <- new_ermod_bin_cov_sel(
    mod = extract_mod(ermod_bin),
    data = df_er_ae_covsel_test,
    var_resp = "AEFLAG",
    var_exposure = "AUCss_1000",
    var_cov_candidates = c("BAGE_10", "BWT_10", "BHBA1C_5", "RACE", "VISC"),
    var_cov = c("BHBA1C_5"),
    var_selected = c("AUCss_1000", "BHBA1C_5"),
    cv_method = c("LOO", "kfold"),
    cvvs = NULL,
    rk = NULL
  )

  test_that("p_direction", {
    expect_equal(p_direction(ermod_bin_cov_sel, as_num = TRUE), 0.9955)
    expect_equal(
      p_direction(ermod_bin_cov_sel, as_num = TRUE, as_p = TRUE),
      0.009
    )
  })
}