fit1 <- stats::lm(Sepal.Length ~ Petal.Width + Species + Sepal.Width, data = iris)
fit2 <- stats::lm(Sepal.Length ~ Petal.Width, data = iris)
fl1 <- flashlight(model = fit1, label = "1", data = iris, y = "Sepal.Length")
fl2 <- flashlight(model = fit2, label = "2", data = iris, y = "Sepal.Length")
fls <- multiflashlight(list(fl1, fl2))

test_that("most_important works ", {
  imp <- light_importance(fl1, seed = 1L)
  expect_equal(most_important(imp, 1L), "Species")
  expect_equal(
    most_important(imp),
    c("Species", "Sepal.Width", "Petal.Width", "Petal.Length")
  )

  imp <- light_importance(fls, seed = 1L)
  expect_equal(most_important(imp, 1L), "Petal.Width")
  expect_equal(
    most_important(imp),
               c("Petal.Width", "Species", "Sepal.Width", "Petal.Length"))
})

test_that("light_importance works", {
  imp <- light_importance(fl2, seed = 1L)
  expect_equal(imp$data$value_[imp$data$variable_ != "Petal.Width"], rep(0, 3))
  expect_true(imp$data$value_[imp$data$variable_ == "Petal.Width"] > 0)
  expect_equal(imp$data$value_[imp$data$variable_ == "Petal.Width"], 0.623468, tolerance = 0.001)
  expect_s3_class(plot(imp), "ggplot")
})

test_that("light_importance reacts on metric", {
  imp <- light_importance(
    fl2, seed = 1L, metric = list(r_squared = MetricsWeighted::r_squared)
  )
  expect_equal(imp$data$value_[imp$data$variable_ != "Petal.Width"], rep(0, 3))
  expect_true(imp$data$value_[imp$data$variable_ == "Petal.Width"] < 0)
  expect_s3_class(plot(imp), "ggplot")
})

test_that("'lower_is_better' works", {
  imp <- light_importance(
    fl2,
    seed = 1L,
    metric = list(r_squared = MetricsWeighted::r_squared),
    lower_is_better = FALSE
  )
  expect_true(imp$data$value_[imp$data$variable_ == "Petal.Width"] > 0)
})

test_that("'by' and 'v' works", {
  fit3 <- stats::lm(Sepal.Length ~ Petal.Width + Species, data = iris)
  fl3 <- flashlight(model = fit3, label = "3", data = iris, y = "Sepal.Length")
  imp <- light_importance(
    fl3, seed = 1L, by = "Species", v = c("Petal.Width", "Species")
  )

  expect_equal(imp$data$value_[imp$data$variable_ == "Species"], rep(0, 3))
  expect_true(all(imp$data$value_[imp$data$variable_ == "Petal.Width"] > 0))
  expect_equal(
    imp$data$value_[imp$data$variable_ == "Petal.Width"],
    c(0.02224045, 0.11677754, 0.11914359),
    tolerance = 0.001
  )
  expect_s3_class(plot(imp), "ggplot")
})

test_that("'w' reacts", {
  fit3 <- stats::lm(Sepal.Length ~ Petal.Width + Species, data = iris)
  fl3 <- flashlight(
    model = fit3, label = "3", data = iris, y = "Sepal.Length", w = "Petal.Length"
  )
  imp <- light_importance(
    fl3, seed = 1L, by = "Species", v = c("Petal.Width", "Species")
  )

  expect_equal(
    imp$data$value_[imp$data$variable_ == "Petal.Width"],
    c(0.02220016, 0.10929391, 0.11598350),
    tolerance = 0.001
  )
  expect_s3_class(plot(imp), "ggplot")
})

test_that("m_repetitions react", {
  imp <- light_importance(
    fl1, seed = 1L, v = c("Petal.Width", "Species"), m_repetitions = 4L
  )
  expect_equal(imp$data$error_, c(0.005633942, 0.017129828), tolerance = 0.001)
  expect_true(all(!is.na(imp$data$error_)))
  expect_s3_class(plot(imp), "ggplot")
})

test_that("m_repetitions react with 'by'", {
  imp <- light_importance(
    fl1, seed = 1L, v = c("Petal.Width", "Species"), m_repetitions = 4L, by = "Species"
  )
  expect_true(is.light(imp))
  expect_true(all(!is.na(imp$data$error_)))
  expect_s3_class(plot(imp), "ggplot")
})

test_that("multiflashlight works", {
  imp <- light_importance(fls, seed = 1L, v = c("Petal.Width", "Species"))
  expect_equal(dim(imp$data), c(4L, 5L))
  expect_equal(
    imp$data$value_,
    c(0.1425708, 0.4369238, 0.5537231, 0.0000000),
    tolerance = 0.001
  )
  expect_s3_class(plot(imp), "ggplot")
})