skip_if_not_installed("nnet")
skip_if_not_installed("MASS")

data("birthwt", package = "MASS")
void <- capture.output({
  m1 <- nnet::multinom(low ~ age + lwt + race + smoke, data = birthwt)
})

test_that("model_info", {
  expect_true(model_info(m1)$is_binomial)
  expect_false(model_info(m1)$is_linear)
})

test_that("n_parameters", {
  expect_identical(n_parameters(m1), 5L)
})

test_that("find_predictors", {
  expect_identical(find_predictors(m1), list(conditional = c("age", "lwt", "race", "smoke")))
  expect_identical(
    find_predictors(m1, flatten = TRUE),
    c("age", "lwt", "race", "smoke")
  )
  expect_null(find_predictors(m1, effects = "random"))
})

test_that("find_response", {
  expect_identical(find_response(m1), "low")
})

test_that("link_inverse", {
  expect_equal(link_inverse(m1)(0.2), plogis(0.2), tolerance = 1e-5)
})

test_that("get_data", {
  expect_identical(nrow(get_data(m1)), 189L)
  expect_identical(
    colnames(get_data(m1)),
    c("low", "age", "lwt", "race", "smoke")
  )
})

test_that("find_formula", {
  expect_length(find_formula(m1), 1)
  expect_equal(
    find_formula(m1),
    list(conditional = as.formula("low ~ age + lwt + race + smoke")),
    ignore_attr = TRUE
  )
})

test_that("find_terms", {
  expect_identical(find_terms(m1), list(
    response = "low",
    conditional = c("age", "lwt", "race", "smoke")
  ))
  expect_identical(
    find_terms(m1, flatten = TRUE),
    c("low", "age", "lwt", "race", "smoke")
  )
})

test_that("n_obs", {
  expect_identical(n_obs(m1), 189L)
})

test_that("linkfun", {
  expect_false(is.null(link_function(m1)))
})

test_that("find_parameters", {
  expect_identical(
    find_parameters(m1),
    list(conditional = c(
      "(Intercept)", "age", "lwt", "race", "smoke"
    ))
  )
  expect_identical(nrow(get_parameters(m1)), 5L)
  expect_identical(
    get_parameters(m1)$Parameter,
    c("(Intercept)", "age", "lwt", "race", "smoke")
  )
})

test_that("find_statistic", {
  expect_identical(find_statistic(m1), "z-statistic")
})

test_that("get_predicted", {
  void <- capture.output({
    # binary outcome
    m1 <- nnet::multinom(low ~ age + lwt + race + smoke, data = birthwt)
    # multinomial outcome
    m2 <- nnet::multinom(ftv ~ age + lwt + race + smoke, data = birthwt)
  })

  # binary outcomes produces an atomic vector
  x <- get_predicted(m1, predict = "classification")
  expect_true(is.atomic(x))
  expect_false(is.null(x))
  expect_null(dim(x))
  expect_true(all(levels(x) %in% c("0", "1")))
  x <- get_predicted(m1, predict = "expectation")
  expect_true(is.atomic(x))
  expect_false(is.null(x))
  expect_null(dim(x))
  x <- get_predicted(m1, predict = NULL, type = "class")
  expect_true(is.atomic(x))
  expect_false(is.null(x))
  expect_null(dim(x))
  expect_true(all(levels(x) %in% c("0", "1")))
  x <- get_predicted(m1, predict = NULL, type = "probs")
  expect_true(is.atomic(x))
  expect_false(is.null(x))
  expect_null(dim(x))

  # multinomial outcomes depends on predict type
  x <- get_predicted(m2, predict = "classification")
  expect_true(is.atomic(x))
  expect_false(is.null(x))
  expect_null(dim(x))
  expect_true(all(levels(x) %in% as.character(0:6)))
  x <- get_predicted(m2, predict = "expectation")
  expect_s3_class(x, "data.frame")
  expect_true(all(c("Row", "Response", "Predicted") %in% colnames(x)))
  x <- get_predicted(m2, predict = NULL, type = "class")
  expect_true(is.atomic(x))
  expect_false(is.null(x))
  expect_null(dim(x))
  expect_true(all(levels(x) %in% as.character(0:6)))
  x <- get_predicted(m2, predict = NULL, type = "probs")
  expect_s3_class(x, "data.frame")
  expect_true(all(c("Row", "Response", "Predicted") %in% colnames(x)))
})