skip_on_cran()

skip_if_not_installed("mmrm")
skip_if_not(packageVersion("insight") > "0.18.8")

data(fev_data, package = "mmrm")

test_that("model_parameters", {
  m1 <- mmrm::mmrm(
    formula = FEV1 ~ RACE + SEX + ARMCD * AVISIT + us(AVISIT | USUBJID),
    data = fev_data
  )

  out1 <- coef(summary(m1))
  out2 <- model_parameters(m1)
  expect_equal(
    as.vector(out1[, "Estimate"]),
    out2$Coefficient,
    tolerance = 1e-4,
    ignore_attr = TRUE
  )
  expect_identical(
    rownames(out1),
    out2$Parameter
  )
  expect_equal(
    as.vector(out1[, "df"]),
    out2$df_error,
    tolerance = 1e-4,
    ignore_attr = TRUE
  )
  expect_equal(
    as.vector(out1[, "Pr(>|t|)"]),
    out2$p,
    tolerance = 1e-4,
    ignore_attr = TRUE
  )
  expect_equal(
    as.vector(out1[, "t value"]),
    out2$t,
    tolerance = 1e-4,
    ignore_attr = TRUE
  )
  expect_equal(
    as.vector(out1[, "Std. Error"]),
    out2$SE,
    tolerance = 1e-4,
    ignore_attr = TRUE
  )
  expect_identical(attributes(out2)$ci_method, "Satterthwaite")
})

test_that("model_parameters", {
  m1 <- mmrm::mmrm(
    formula = FEV1 ~ RACE + SEX + ARMCD * AVISIT + us(AVISIT | USUBJID),
    data = fev_data,
    method = "Kenward-Roger"
  )

  out1 <- coef(summary(m1))
  out2 <- model_parameters(m1)

  expect_equal(
    as.vector(out1[, "Estimate"]),
    out2$Coefficient,
    tolerance = 1e-4,
    ignore_attr = TRUE
  )
  expect_identical(
    rownames(out1),
    out2$Parameter
  )
  expect_equal(
    as.vector(out1[, "df"]),
    out2$df_error,
    tolerance = 1e-4,
    ignore_attr = TRUE
  )
  expect_equal(
    as.vector(out1[, "Pr(>|t|)"]),
    out2$p,
    tolerance = 1e-4,
    ignore_attr = TRUE
  )
  expect_equal(
    as.vector(out1[, "t value"]),
    out2$t,
    tolerance = 1e-4,
    ignore_attr = TRUE
  )
  expect_equal(
    as.vector(out1[, "Std. Error"]),
    out2$SE,
    tolerance = 1e-4,
    ignore_attr = TRUE
  )
  expect_identical(attributes(out2)$ci_method, "Kenward")
})