tests/testthat/test-simpA.measures.R

In CondCopulas: Estimation and Inference for Conditional Copula Models

test_that("measures_nonsimplifyingness_NP does not work for non-valid input", {

  X1 = c(1,2)
  X2 = 3
  Z = 4

  expect_error({
    measures_nonsimplifyingness_NP(X1 = X1, X2 = X2, Z = Z, h = 1) },
    class = "DifferentLengthsError")

  X1 = matrix(c(1,2), ncol = 2)

  expect_error({
    measures_nonsimplifyingness_NP(X1 = X1, X2 = X2, Z = Z, h = 1) },
    class = "WrongDimensionError")

  X1 = 2
  Z = matrix(c(1,2), ncol = 2)

  expect_error({
    measures_nonsimplifyingness_NP(X1 = X1, X2 = X2, Z = Z, h = 1) },
    class = "WrongDimensionError")

  Z = 2

  # We now give invalid measures

  expect_error({
    measures_nonsimplifyingness_NP(X1 = X1, X2 = X2, Z = Z, h = 1, measures = c()) },
    class = "ZeroLengthError")

  expect_error({
    measures_nonsimplifyingness_NP(X1 = X1, X2 = X2, Z = Z, h = 1, measures = "aaaa") },
    class = "UnknownMeasureNameError")

})


test_that("measures_nonsimplifyingness_NP works for valid input", {

  X1 = rnorm(100)
  X2 = rnorm(100)
  Z = rnorm(100)

  # It works by default
  expect_no_error({
    measures_nonsimplifyingness_NP(X1 = X1, X2 = X2, Z = Z, h = 0.1) })

  # It works for one of the measures
  expect_no_error({
    measures_nonsimplifyingness_NP(X1 = X1, X2 = X2, Z = Z, h = 1,
                                   measures = "T1_CvM_Cs3") })

  # It works for two measures
  expect_no_error({
    measures_nonsimplifyingness_NP(X1 = X1, X2 = X2, Z = Z, h = 1,
                                   measures = c("T1_CvM_Cs3", "T1_CvM_Cs4")) })
})


test_that("measures_nonsimplifyingness_NP works for several bandwidths at the same time", {

  X1 = rnorm(100)
  X2 = rnorm(100)
  Z = rnorm(100)

  measures = measures_nonsimplifyingness_NP(X1 = X1, X2 = X2, Z = Z, h = c(0.1, 1),
                                            verbose = 1)

  expect_true("h" %in% colnames(measures))
  expect_equal(sort(unique(measures$h)), c(0.1, 1) )

  measures_0_1 = measures_nonsimplifyingness_NP(X1 = X1, X2 = X2, Z = Z, h = 0.1,
                                                verbose = 1)

  expect_identical(measures[which(measures$h == 0.1), c("measure", "value")],
                   measures_0_1[, c("measure", "value")])
})


test_that("measures_nonsimplifyingness_NP does not work if truncVal is invalid", {

  n = 50
  X1 = rnorm(n)
  X2 = rnorm(n)
  Z = rnorm(n)

  expect_error(
    {measures = measures_nonsimplifyingness_NP(X1 = X1, X2 = X2, Z = Z, h = c(0.5, 1),
                                               truncVal = -1)},
  )

  expect_error(
    {measures = measures_nonsimplifyingness_NP(X1 = X1, X2 = X2, Z = Z, h = c(0.5, 1),
                                               truncVal = 0.5)},
  )
})