Merge pull request #43 from tidymodels/test-arrow

test on arrow tables

tests/testthat/test-step_adasyn.R

@@ -127,3 +127,25 @@ test_that("duckdb - step_adasyn works", {
 
   DBI::dbDisconnect(con)
 })
+
+test_that("arrow - step_adasyn works", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("themis")
+  skip_if_not_installed("arrow")
+
+  mtcars_adasyn <- dplyr::as_tibble(mtcars)
+  mtcars_adasyn$vs <- as.factor(mtcars$vs)
+
+  rec <- recipes::recipe(mpg ~ ., data = mtcars_adasyn) %>%
+    themis::step_adasyn(vs, skip = TRUE) %>%
+    recipes::prep()
+
+  res <- dplyr::mutate(mtcars_adasyn, !!!orbital_inline(orbital(rec)))
+
+  mtcars_tbl <- arrow::as_arrow_table(mtcars_adasyn)
+
+  res_new <- dplyr::mutate(mtcars_tbl, !!!orbital_inline(orbital(rec))) %>%
+    dplyr::collect()
+
+  expect_equal(res_new, res)
+})

tests/testthat/test-step_bin2factor.R

@@ -119,3 +119,23 @@ test_that("duckdb - step_bin2factor works", {
 
   DBI::dbDisconnect(con)
 })
+
+test_that("arrow - step_bin2factor works", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("arrow")
+
+  mtcars <- dplyr::as_tibble(mtcars)
+
+  rec <- recipes::recipe(mpg ~ ., data = mtcars) %>%
+    recipes::step_bin2factor(vs, am) %>%
+    recipes::prep()
+
+  res <- dplyr::mutate(mtcars, !!!orbital_inline(orbital(rec)))
+
+  mtcars_tbl <- arrow::as_arrow_table(mtcars)
+
+  res_new <- dplyr::mutate(mtcars_tbl, !!!orbital_inline(orbital(rec))) %>%
+    dplyr::collect()
+
+  expect_equal(res_new, res)
+})

tests/testthat/test-step_boxcox.R

@@ -117,3 +117,23 @@ test_that("duckdb - step_BoxCox works", {
 
   DBI::dbDisconnect(con)
 })
+
+test_that("arrow - step_BoxCox works", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("arrow")
+
+  mtcars <- dplyr::as_tibble(mtcars)
+
+  rec <- recipes::recipe(~ ., data = mtcars) %>%
+    recipes::step_BoxCox(mpg, disp) %>%
+    recipes::prep()
+
+  res <- dplyr::mutate(mtcars, !!!orbital_inline(orbital(rec)))
+
+  mtcars_tbl <- arrow::as_arrow_table(mtcars)
+
+  res_new <- dplyr::mutate(mtcars_tbl, !!!orbital_inline(orbital(rec))) %>%
+    dplyr::collect()
+
+  expect_equal(res_new, res)
+})

tests/testthat/test-step_bsmote.R

@@ -127,3 +127,25 @@ test_that("duckdb - step_bsmote works", {
 
   DBI::dbDisconnect(con)
 })
+
+test_that("arrow - step_bsmote works", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("themis")
+  skip_if_not_installed("arrow")
+
+  mtcars_bsmote <- dplyr::as_tibble(mtcars)
+  mtcars_bsmote$vs <- as.factor(mtcars$vs)
+
+  rec <- recipes::recipe(mpg ~ ., data = mtcars_bsmote) %>%
+    themis::step_bsmote(vs, skip = TRUE) %>%
+    recipes::prep()
+
+  res <- dplyr::mutate(mtcars_bsmote, !!!orbital_inline(orbital(rec)))
+
+  mtcars_tbl <- arrow::as_arrow_table(mtcars_bsmote)
+
+  res_new <- dplyr::mutate(mtcars_tbl, !!!orbital_inline(orbital(rec))) %>%
+    dplyr::collect()
+
+  expect_equal(res_new, res)
+})

tests/testthat/test-step_center.R

@@ -117,3 +117,23 @@ test_that("duckdb - step_center works", {
 
   DBI::dbDisconnect(con)
 })
+
+test_that("arrow - step_center works", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("arrow")
+
+  mtcars <- dplyr::as_tibble(mtcars)
+
+  rec <- recipes::recipe(mpg ~ ., data = mtcars) %>%
+    recipes::step_center(recipes::all_predictors()) %>%
+    recipes::prep()
+
+  res <- dplyr::mutate(mtcars, !!!orbital_inline(orbital(rec)))
+
+  mtcars_tbl <- arrow::as_arrow_table(mtcars)
+
+  res_new <- dplyr::mutate(mtcars_tbl, !!!orbital_inline(orbital(rec))) %>%
+    dplyr::collect()
+
+  expect_equal(res_new, res)
+})

tests/testthat/test-step_corr.R

@@ -42,4 +42,9 @@ test_that("SQLite - step_corr works", {
 test_that("duckdb - step_corr works", {
   # Isn't needed as `step_corr()` doesn't produce code
   expect_true(TRUE)
+})
+
+test_that("arrow - step_corr works", {
+  # Isn't needed as `step_corr()` doesn't produce code
+  expect_true(TRUE)
 })

tests/testthat/test-step_discretize.R

@@ -176,3 +176,26 @@ test_that("duckdb - step_discretize works", {
 
   DBI::dbDisconnect(con)
 })
+
+test_that("arrow - step_discretize works", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("arrow")
+
+  mtcars_discretize <- dplyr::as_tibble(mtcars)
+  mtcars_discretize[1, ] <- NA
+
+  suppressWarnings(
+    rec <- recipes::recipe(mpg ~ ., data = mtcars_discretize) %>%
+      recipes::step_discretize(mpg, disp, min_unique = 4) %>%
+      recipes::prep()
+  )
+
+  res <- dplyr::mutate(mtcars_discretize, !!!orbital_inline(orbital(rec)))
+
+  mtcars_tbl <- arrow::as_arrow_table(mtcars_discretize)
+
+  res_new <- dplyr::mutate(mtcars_tbl, !!!orbital_inline(orbital(rec))) %>%
+    dplyr::collect()
+
+  expect_equal(res_new, res)
+})

tests/testthat/test-step_downsample.R

@@ -127,3 +127,25 @@ test_that("duckdb - step_downsample works", {
 
   DBI::dbDisconnect(con)
 })
+
+test_that("arrow - step_downsample works", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("themis")
+  skip_if_not_installed("arrow")
+
+  mtcars_downsample <- dplyr::as_tibble(mtcars)
+  mtcars_downsample$vs <- as.factor(mtcars$vs)
+
+  rec <- recipes::recipe(mpg ~ ., data = mtcars_downsample) %>%
+    themis::step_downsample(vs, skip = TRUE) %>%
+    recipes::prep()
+
+  res <- dplyr::mutate(mtcars_downsample, !!!orbital_inline(orbital(rec)))
+
+  mtcars_tbl <- arrow::as_arrow_table(mtcars_downsample)
+
+  res_new <- dplyr::mutate(mtcars_tbl, !!!orbital_inline(orbital(rec))) %>%
+    dplyr::collect()
+
+  expect_equal(res_new, res)
+})

tests/testthat/test-step_dummy.R

@@ -153,4 +153,27 @@ test_that("duckdb - step_dummy works", {
   expect_equal(res_new, exp)
 
   DBI::dbDisconnect(con)
+})
+
+test_that("arrow - step_dummy works", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("arrow")
+
+  mtcars1 <- dplyr::as_tibble(mtcars)
+
+  mtcars1$gear <- as.character(mtcars1$gear)
+  mtcars1$carb <- as.character(mtcars1$carb)
+
+  rec <- recipes::recipe(mpg ~ ., data = mtcars1) %>%
+    recipes::step_dummy(recipes::all_nominal_predictors()) %>%
+    recipes::prep()
+
+  exp <- dplyr::mutate(mtcars1, !!!orbital_inline(orbital(rec)))
+
+  mtcars_tbl <- arrow::as_arrow_table(mtcars1)
+
+  res_new <- dplyr::mutate(mtcars_tbl, !!!orbital_inline(orbital(rec))) %>%
+    dplyr::collect()
+
+  expect_equal(res_new, exp)
 })

tests/testthat/test-step_filter_missing.R

@@ -41,4 +41,9 @@ test_that("SQLite - step_filter_missing works", {
 test_that("duckdb - step_filter_missing works", {
   # Isn't needed as `step_filter_missing()` doesn't produce code
   expect_true(TRUE)
+})
+
+test_that("arrow - step_filter_missing works", {
+  # Isn't needed as `step_filter_missing()` doesn't produce code
+  expect_true(TRUE)
 })

tests/testthat/test-step_impute_mean.R

@@ -122,3 +122,24 @@ test_that("duckdb - step_impute_mean works", {
 
   DBI::dbDisconnect(con)
 })
+
+test_that("arrow - step_impute_mean works", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("arrow")
+
+  mtcars_impute_mean <- dplyr::as_tibble(mtcars)
+  mtcars_impute_mean[2:4, ] <- NA
+
+  rec <- recipes::recipe(mpg ~ ., data = mtcars_impute_mean) %>%
+    recipes::step_impute_mean(recipes::all_predictors()) %>%
+    recipes::prep()
+
+  res <- dplyr::mutate(mtcars_impute_mean, !!!orbital_inline(orbital(rec)))
+
+  mtcars_tbl <- arrow::as_arrow_table(mtcars_impute_mean)
+
+  res_new <- dplyr::mutate(mtcars_tbl, !!!orbital_inline(orbital(rec))) %>%
+    dplyr::collect()
+
+  expect_equal(res_new, res)
+})

tests/testthat/test-step_impute_median.R

@@ -123,3 +123,24 @@ test_that("duckdb - step_impute_median works", {
 
   DBI::dbDisconnect(con)
 })
+
+test_that("arrow - step_impute_median works", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("arrow")
+
+  mtcars_impute_median <- dplyr::as_tibble(mtcars)
+  mtcars_impute_median[2:4, ] <- NA
+
+  rec <- recipes::recipe(mpg ~ ., data = mtcars_impute_median) %>%
+    recipes::step_impute_median(recipes::all_predictors()) %>%
+    recipes::prep()
+
+  res <- dplyr::mutate(mtcars_impute_median, !!!orbital_inline(orbital(rec)))
+
+  mtcars_tbl <- arrow::as_arrow_table(mtcars_impute_median)
+
+  res_new <- dplyr::mutate(mtcars_tbl, !!!orbital_inline(orbital(rec))) %>%
+    dplyr::collect()
+
+  expect_equal(res_new, res)
+})

tests/testthat/test-step_impute_mode.R

@@ -135,3 +135,26 @@ test_that("duckdb - step_impute_mode works", {
 
   DBI::dbDisconnect(con)
 })
+
+test_that("arrow - step_impute_mode works", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("arrow")
+
+  mtcars_impute_mode <- dplyr::as_tibble(mtcars)
+  mtcars_impute_mode$gear <- letters[mtcars$gear]
+  mtcars_impute_mode$carb <- letters[mtcars$carb]
+  mtcars_impute_mode[2:4, ] <- NA
+
+  rec <- recipes::recipe(mpg ~ ., data = mtcars_impute_mode) %>%
+    recipes::step_impute_mode(recipes::all_nominal_predictors()) %>%
+    recipes::prep(strings_as_factors = FALSE)
+
+  res <- dplyr::mutate(mtcars_impute_mode, !!!orbital_inline(orbital(rec)))
+
+  mtcars_tbl <- arrow::as_arrow_table(mtcars_impute_mode)
+
+  res_new <- dplyr::mutate(mtcars_tbl, !!!orbital_inline(orbital(rec))) %>%
+    dplyr::collect()
+
+  expect_equal(res_new, res)
+})

tests/testthat/test-step_indicate_na.R

@@ -126,3 +126,24 @@ test_that("duckdb - step_indicate_na works", {
 
   DBI::dbDisconnect(con)
 })
+
+test_that("arrow - step_indicate_na works", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("arrow")
+
+  mtcars_indicate_na <- dplyr::as_tibble(mtcars)
+  mtcars_indicate_na[2:4, ] <- NA
+
+  rec <- recipes::recipe(mpg ~ ., data = mtcars_indicate_na) %>%
+    recipes::step_indicate_na(recipes::all_predictors()) %>%
+    recipes::prep()
+
+  res <- dplyr::mutate(mtcars_indicate_na, !!!orbital_inline(orbital(rec)))
+
+  mtcars_tbl <- arrow::as_arrow_table(mtcars_indicate_na)
+
+  res_new <- dplyr::mutate(mtcars_tbl, !!!orbital_inline(orbital(rec))) %>%
+    dplyr::collect()
+
+  expect_equal(res_new, res)
+})

tests/testthat/test-step_intercept.R

@@ -107,3 +107,23 @@ test_that("duckdb - step_intercept works", {
 
   DBI::dbDisconnect(con)
 })
+
+test_that("arrow - step_intercept works", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("arrow")
+
+  mtcars <- dplyr::as_tibble(mtcars)
+
+  rec <- recipes::recipe(mpg ~ ., data = mtcars) %>%
+    recipes::step_intercept(value = 5) %>%
+    recipes::prep()
+
+  res <- dplyr::mutate(mtcars, !!!orbital_inline(orbital(rec)))
+
+  mtcars_tbl <- arrow::as_arrow_table(mtcars)
+
+  res_new <- dplyr::mutate(mtcars_tbl, !!!orbital_inline(orbital(rec))) %>%
+    dplyr::collect()
+
+  expect_equal(res_new, res)
+})