tests/testthat/test-impute_roll.R

In recipes: Preprocessing and Feature Engineering Steps for Modeling

library(testthat)
library(recipes)

set.seed(145)
example_data <-
  data.frame(
    day = lubridate::ymd("2012-06-07") + lubridate::days(1:12),
    x1 = round(runif(12), 2),
    x2 = round(runif(12), 2),
    x3 = round(runif(12), 2)
  )
example_data$x1[c(1, 5, 6)] <- NA
example_data$x2[c(1:4, 10)] <- NA
example_data <- as_tibble(example_data)

test_that("imputation values with 7-pt median", {
  seven_pt <- recipe(~., data = example_data) %>%
    update_role(day, new_role = "time_index") %>%
    step_impute_roll(all_predictors(), window = 7, id = "") %>%
    prep(training = example_data)

  seven_pt_exp <- example_data
  seven_pt_exp$x1[1] <- median(seven_pt_exp$x1[1:7], na.rm = TRUE)
  seven_pt_exp$x1[5] <- median(seven_pt_exp$x1[2:8], na.rm = TRUE)
  seven_pt_exp$x1[6] <- median(seven_pt_exp$x1[3:9], na.rm = TRUE)
  seven_pt_exp$x2[1] <- median(seven_pt_exp$x2[1:7], na.rm = TRUE)
  seven_pt_exp$x2[2] <- median(seven_pt_exp$x2[1:7], na.rm = TRUE)
  seven_pt_exp$x2[3] <- median(seven_pt_exp$x2[1:7], na.rm = TRUE)
  seven_pt_exp$x2[4] <- median(seven_pt_exp$x2[1:7], na.rm = TRUE)
  seven_pt_exp$x2[10] <- median(seven_pt_exp$x2[6:12], na.rm = TRUE)

  expect_equal(seven_pt_exp, bake(seven_pt, new_data = NULL))

  seven_pt_tidy_tr <-
    tibble(
      terms = paste0("x", 1:3),
      window = rep(7L, 3),
      id = ""
    )
  expect_equal(seven_pt_tidy_tr, tidy(seven_pt, number = 1))
})

test_that("imputation values with 3-pt mean", {
  three_pt <- recipe(~., data = example_data) %>%
    update_role(day, new_role = "time_index") %>%
    step_impute_roll(all_predictors(), window = 3, id = "") %>%
    prep(training = example_data)


  three_pt_exp <- example_data
  three_pt_exp$x1[1] <- mean(three_pt_exp$x1[1:3], na.rm = TRUE)
  three_pt_exp$x1[5] <- mean(three_pt_exp$x1[4:6], na.rm = TRUE)
  three_pt_exp$x1[6] <- mean(example_data$x1[5:7], na.rm = TRUE)
  three_pt_exp$x2[1] <- NA
  three_pt_exp$x2[2] <- NA
  three_pt_exp$x2[3] <- NA
  three_pt_exp$x2[4] <- mean(three_pt_exp$x2[3:5], na.rm = TRUE)
  three_pt_exp$x2[10] <- mean(three_pt_exp$x2[9:11], na.rm = TRUE)

  expect_equal(three_pt_exp, bake(three_pt, new_data = NULL))

  three_pt_tidy_tr <-
    tibble(
      terms = paste0("x", 1:3),
      window = rep(3L, 3),
      id = ""
    )
  expect_equal(three_pt_tidy_tr, tidy(three_pt, number = 1))
})


test_that("bad args", {
  expect_snapshot(error = TRUE,
    recipe(~., data = example_data) %>%
      step_impute_roll(all_predictors(), window = 3) %>%
      prep(training = example_data)
  )

  expect_snapshot(error = TRUE,
    recipe(~., data = example_data) %>%
      update_role(day, new_role = "time_index") %>%
      step_impute_roll(all_predictors(), window = 4) %>%
      prep(training = example_data)
  )

  example_data$x4 <- 1:12
  expect_snapshot(error = TRUE,
    recipe(~., data = example_data) %>%
      update_role(day, new_role = "time_index") %>%
      step_impute_roll(all_predictors(), window = 3) %>%
      prep(training = example_data)
  )
})

test_that("Deprecation warning", {
  expect_snapshot(error = TRUE,
    recipe(~ ., data = mtcars) %>%
      step_rollimpute()
  )
})

test_that("tunable", {
  rec <-
    recipe(~., data = iris) %>%
    step_impute_roll(all_predictors(), outcome = "Species")
  rec_param <- tunable.step_impute_roll(rec$steps[[1]])
  expect_equal(rec_param$name, c("statistic", "window"))
  expect_true(all(rec_param$source == "recipe"))
  expect_true(is.list(rec_param$call_info))
  expect_equal(nrow(rec_param), 2)
  expect_equal(
    names(rec_param),
    c("name", "call_info", "source", "component", "component_id")
  )
})

# Infrastructure ---------------------------------------------------------------

test_that("bake method errors when needed non-standard role columns are missing", {
  seven_pt <- recipe(~., data = example_data) %>%
    update_role(day, new_role = "time_index") %>%
    step_impute_roll(x1, window = 7) %>%
    update_role(x1, new_role = "potato") %>%
    update_role_requirements(role = "potato", bake = FALSE) %>%
    prep(training = example_data)

  expect_error(bake(seven_pt, new_data = example_data[, c(-2)]),
               class = "new_data_missing_column")
})

test_that("empty printing", {
  rec <- recipe(mpg ~ ., mtcars)
  rec <- step_impute_roll(rec)

  expect_snapshot(rec)

  rec <- prep(rec, mtcars)

  expect_snapshot(rec)
})

test_that("empty selection prep/bake is a no-op", {
  rec1 <- recipe(mpg ~ ., mtcars)
  rec2 <- step_impute_roll(rec1)

  rec1 <- prep(rec1, mtcars)
  rec2 <- prep(rec2, mtcars)

  baked1 <- bake(rec1, mtcars)
  baked2 <- bake(rec2, mtcars)

  expect_identical(baked1, baked2)
})

test_that("empty selection tidy method works", {
  rec <- recipe(mpg ~ ., mtcars)
  rec <- step_impute_roll(rec)

  expect <- tibble(terms = character(), window = integer(), id = character())

  expect_identical(tidy(rec, number = 1), expect)

  rec <- prep(rec, mtcars)

  expect_identical(tidy(rec, number = 1), expect)
})

test_that("printing", {
  rec <- recipe(~., data = example_data) %>%
    update_role(day, new_role = "time_index") %>%
    step_impute_roll(all_predictors(), window = 7)

  expect_snapshot(print(rec))
  expect_snapshot(prep(rec))
})

test_that("tunable is setup to work with extract_parameter_set_dials", {
  skip_if_not_installed("dials")
  rec <- recipe(~., data = mtcars) %>%
    step_impute_roll(
      all_predictors(),
      statistic = hardhat::tune(), window = hardhat::tune()
    )

  params <- extract_parameter_set_dials(rec)

  expect_s3_class(params, "parameters")
  expect_identical(nrow(params), 2L)
})