tests/testthat/test-unnest_signals.R

In mkerschbaumer/mbte: Model-based trend estimation

context("unnest_signals")

raw_tbl_mbte <- gen_raw_tbl_mbte()

# mbte_unnest_signals() should check integrity of input parameter
test_input_not_tbl_mbte(mbte_unnest_signals)

# test signal column not present or malformatted
test_signal_col_np_mf(mbte_unnest_signals, raw_tbl_mbte)

# mbte_unnest_signals() should invert mbte_nest_signals()
test_that("inversion of mbte_nest_signals", {
  # make sure unnesting also works if the signal variable is not named
  # "signal" (default)
  tbl <- new_tbl_mbte(raw_signals, t, value, signal = signal_var)

  nested <- expect_silent(mbte_nest_signals(tbl, mv))
  unnested <- expect_silent(mbte_unnest_signals(nested))

  # reorder columns (since mbte_unnest_signals() always puts time- and value-
  # columns after "descriptive" columns like `mv`)
  reordered <- dplyr::select(tbl, mv, dplyr::everything())

  # make sure inversion has been successful
  expect_tbl_mbte_equal(unnested, reordered)
})

test_that("positive test with `fits`-column", {
  # perform fitting (add `fits`-column)
  fitted <- mbte_fit(filtered_signals, lm = lm(value ~ t, .signal))

  # unnesting of tbl_mbte with `fits`-column should yield the same results as
  # unnesting a tbl_mbte without `fits`-column
  fitted_unnested <- mbte_unnest_signals(fitted)
  original_unnested <- mbte_unnest_signals(filtered_signals)
  expect_tbl_mbte_equal(fitted_unnested, original_unnested)
})