Themis steps

DESCRIPTION

@@ -29,6 +29,7 @@ Suggests:
     sparklyr,
     testthat (>= 3.0.0),
     tidypredict,
+    themis,
     workflows
 Config/testthat/edition: 3
 Encoding: UTF-8

NAMESPACE

@@ -6,8 +6,11 @@ S3method(orbital,model_fit)
 S3method(orbital,model_spec)
 S3method(orbital,recipe)
 S3method(orbital,step_BoxCox)
+S3method(orbital,step_adasyn)
+S3method(orbital,step_bsmote)
 S3method(orbital,step_center)
 S3method(orbital,step_corr)
+S3method(orbital,step_downsample)
 S3method(orbital,step_dummy)
 S3method(orbital,step_filter_missing)
 S3method(orbital,step_impute_mean)
@@ -20,6 +23,7 @@ S3method(orbital,step_lag)
 S3method(orbital,step_lincomb)
 S3method(orbital,step_log)
 S3method(orbital,step_mutate)
+S3method(orbital,step_nearmiss)
 S3method(orbital,step_normalize)
 S3method(orbital,step_novel)
 S3method(orbital,step_nzv)
@@ -29,10 +33,15 @@ S3method(orbital,step_range)
 S3method(orbital,step_ratio)
 S3method(orbital,step_rename)
 S3method(orbital,step_rm)
+S3method(orbital,step_rose)
 S3method(orbital,step_scale)
 S3method(orbital,step_select)
+S3method(orbital,step_smote)
+S3method(orbital,step_smotenc)
 S3method(orbital,step_sqrt)
+S3method(orbital,step_tomek)
 S3method(orbital,step_unknown)
+S3method(orbital,step_upsample)
 S3method(orbital,step_zv)
 S3method(orbital,workflow)
 S3method(predict,orbital_class)

NEWS.md

@@ -36,6 +36,8 @@
 
 * Support for `step_rename()` has been added. (#17)
 
+* Support for `step_upsample()`, `step_smote()`, `step_smotenc()`, `step_bsmote()`, `step_adasyn()`, `step_rose()`, `step_downsample()`, `step_nearmiss()`, and `step_tomek()` has been added. (#21)
+
 * `orbital()` now works on `tune::last_fit()` objects. (#13)
 
 * `orbital_predict()` has been removed and replaced with the more idiomatic `predict()` method. (#10)

R/step_adasyn.R

@@ -0,0 +1,7 @@
+#' @export
+orbital.step_adasyn <- function(x, all_vars, ...) {
+  cli::cli_abort(
+    "{.fn orbital} method doesn't work for {.fn step_adasyn} when \\
+    {.arg skip} is {.code FALSE}."
+  )
+}

R/step_bsmote.R

@@ -0,0 +1,7 @@
+#' @export
+orbital.step_bsmote <- function(x, all_vars, ...) {
+  cli::cli_abort(
+    "{.fn orbital} method doesn't work for {.fn step_bsmote} when \\
+    {.arg skip} is {.code FALSE}."
+  )
+}

R/step_downsample.R

@@ -0,0 +1,7 @@
+#' @export
+orbital.step_downsample <- function(x, all_vars, ...) {
+  cli::cli_abort(
+    "{.fn orbital} method doesn't work for {.fn step_downsample} when \\
+    {.arg skip} is {.code FALSE}."
+  )
+}

R/step_nearmiss.R

@@ -0,0 +1,7 @@
+#' @export
+orbital.step_nearmiss <- function(x, all_vars, ...) {
+  cli::cli_abort(
+    "{.fn orbital} method doesn't work for {.fn step_nearmiss} when \\
+    {.arg skip} is {.code FALSE}."
+  )
+}

R/step_rose.R

@@ -0,0 +1,7 @@
+#' @export
+orbital.step_rose <- function(x, all_vars, ...) {
+  cli::cli_abort(
+    "{.fn orbital} method doesn't work for {.fn step_rose} when \\
+    {.arg skip} is {.code FALSE}."
+  )
+}

R/step_smote.R

@@ -0,0 +1,7 @@
+#' @export
+orbital.step_smote <- function(x, all_vars, ...) {
+  cli::cli_abort(
+    "{.fn orbital} method doesn't work for {.fn step_smote} when \\
+    {.arg skip} is {.code FALSE}."
+  )
+}

R/step_smotenc.R

@@ -0,0 +1,7 @@
+#' @export
+orbital.step_smotenc <- function(x, all_vars, ...) {
+  cli::cli_abort(
+    "{.fn orbital} method doesn't work for {.fn step_smotenc} when \\
+    {.arg skip} is {.code FALSE}."
+  )
+}

R/step_tomek.R

@@ -0,0 +1,7 @@
+#' @export
+orbital.step_tomek <- function(x, all_vars, ...) {
+  cli::cli_abort(
+    "{.fn orbital} method doesn't work for {.fn step_tomek} when \\
+    {.arg skip} is {.code FALSE}."
+  )
+}

R/step_upsample.R

@@ -0,0 +1,7 @@
+#' @export
+orbital.step_upsample <- function(x, all_vars, ...) {
+  cli::cli_abort(
+    "{.fn orbital} method doesn't work for {.fn step_upsample} when \\
+    {.arg skip} is {.code FALSE}."
+  )
+}

tests/testthat/_snaps/step_adasyn.md

@@ -0,0 +1,8 @@
+# step_bsmote errors with skip = FALSE
+
+    Code
+      orbital(rec)
+    Condition
+      Error in `orbital()`:
+      ! `orbital()` method doesn't work for `step_bsmote()` when `skip` is `FALSE`.
+

tests/testthat/_snaps/step_bsmote.md

@@ -0,0 +1,8 @@
+# step_bsmote errors with skip = FALSE
+
+    Code
+      orbital(rec)
+    Condition
+      Error in `orbital()`:
+      ! `orbital()` method doesn't work for `step_bsmote()` when `skip` is `FALSE`.
+

tests/testthat/_snaps/step_downsample.md

@@ -0,0 +1,8 @@
+# step_downsample errors with skip = FALSE
+
+    Code
+      orbital(rec)
+    Condition
+      Error in `orbital()`:
+      ! `orbital()` method doesn't work for `step_downsample()` when `skip` is `FALSE`.
+

tests/testthat/_snaps/step_nearmiss.md

@@ -0,0 +1,8 @@
+# step_nearmiss errors with skip = FALSE
+
+    Code
+      orbital(rec)
+    Condition
+      Error in `orbital()`:
+      ! `orbital()` method doesn't work for `step_nearmiss()` when `skip` is `FALSE`.
+

tests/testthat/_snaps/step_rose.md

@@ -0,0 +1,8 @@
+# step_rose errors with skip = FALSE
+
+    Code
+      orbital(rec)
+    Condition
+      Error in `orbital()`:
+      ! `orbital()` method doesn't work for `step_rose()` when `skip` is `FALSE`.
+

tests/testthat/_snaps/step_smote.md

@@ -0,0 +1,8 @@
+# step_smote errors with skip = FALSE
+
+    Code
+      orbital(rec)
+    Condition
+      Error in `orbital()`:
+      ! `orbital()` method doesn't work for `step_smote()` when `skip` is `FALSE`.
+

tests/testthat/_snaps/step_smotenc.md

@@ -0,0 +1,8 @@
+# step_smotenc errors with skip = FALSE
+
+    Code
+      orbital(rec)
+    Condition
+      Error in `orbital()`:
+      ! `orbital()` method doesn't work for `step_smotenc()` when `skip` is `FALSE`.
+

tests/testthat/_snaps/step_tomek.md

@@ -0,0 +1,8 @@
+# step_tomek errors with skip = FALSE
+
+    Code
+      orbital(rec)
+    Condition
+      Error in `orbital()`:
+      ! `orbital()` method doesn't work for `step_tomek()` when `skip` is `FALSE`.
+

tests/testthat/_snaps/step_upsample.md

@@ -0,0 +1,8 @@
+# step_upsample errors with skip = FALSE
+
+    Code
+      orbital(rec)
+    Condition
+      Error in `orbital()`:
+      ! `orbital()` method doesn't work for `step_upsample()` when `skip` is `FALSE`.
+

tests/testthat/test-step_adasyn.R

@@ -0,0 +1,60 @@
+test_that("step_adasyn works", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("themis")
+
+  mtcars <- dplyr::as_tibble(mtcars)
+  mtcars$vs <- as.factor(mtcars$vs)
+
+  rec <- recipes::recipe(mpg ~ ., data = mtcars) %>%
+    themis::step_adasyn(vs, skip = TRUE) %>%
+    recipes::prep()
+
+  res <- dplyr::mutate(mtcars, !!!orbital_inline(orbital(rec)))
+
+  exp <- recipes::bake(rec, new_data = mtcars)
+  exp <- exp[names(res)]
+
+  expect_equal(res, exp)
+})
+
+test_that("step_adasyn errors with skip = FALSE", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("themis")
+
+  mtcars <- dplyr::as_tibble(mtcars)
+  mtcars$vs <- as.factor(mtcars$vs)
+
+  rec <- recipes::recipe(mpg ~ ., data = mtcars) %>%
+    themis::step_adasyn(vs, skip = FALSE) %>%
+    recipes::prep()
+
+  expect_snapshot(
+    error = TRUE,
+    orbital(rec)
+  )
+})
+
+test_that("spark - step_adasyn works", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("themis")
+  skip_if_not_installed("sparklyr")
+  skip_if(is.na(testthat_spark_env_version()))
+
+  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)))
+  res$vs <- as.character(res$vs)
+
+  sc <- testthat_spark_connection()
+  mtcars_tbl <- testthat_tbl("mtcars_adasyn")
+
+  res_spark <- dplyr::mutate(mtcars_tbl, !!!orbital_inline(orbital(rec))) %>%
+    dplyr::collect()
+
+  expect_equal(res_spark, res)
+})

tests/testthat/test-step_bsmote.R

@@ -0,0 +1,60 @@
+test_that("step_bsmote works", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("themis")
+
+  mtcars <- dplyr::as_tibble(mtcars)
+  mtcars$vs <- as.factor(mtcars$vs)
+
+  rec <- recipes::recipe(mpg ~ ., data = mtcars) %>%
+    themis::step_bsmote(vs, skip = TRUE) %>%
+    recipes::prep()
+
+  res <- dplyr::mutate(mtcars, !!!orbital_inline(orbital(rec)))
+
+  exp <- recipes::bake(rec, new_data = mtcars)
+  exp <- exp[names(res)]
+
+  expect_equal(res, exp)
+})
+
+test_that("step_bsmote errors with skip = FALSE", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("themis")
+
+  mtcars <- dplyr::as_tibble(mtcars)
+  mtcars$vs <- as.factor(mtcars$vs)
+
+  rec <- recipes::recipe(mpg ~ ., data = mtcars) %>%
+    themis::step_bsmote(vs, skip = FALSE) %>%
+    recipes::prep()
+
+  expect_snapshot(
+    error = TRUE,
+    orbital(rec)
+  )
+})
+
+test_that("spark - step_bsmote works", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("themis")
+  skip_if_not_installed("sparklyr")
+  skip_if(is.na(testthat_spark_env_version()))
+
+  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)))
+  res$vs <- as.character(res$vs)
+
+  sc <- testthat_spark_connection()
+  mtcars_tbl <- testthat_tbl("mtcars_bsmote")
+
+  res_spark <- dplyr::mutate(mtcars_tbl, !!!orbital_inline(orbital(rec))) %>%
+    dplyr::collect()
+
+  expect_equal(res_spark, res)
+})

tests/testthat/test-step_downsample.R

@@ -0,0 +1,60 @@
+test_that("step_downsample works", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("themis")
+
+  mtcars <- dplyr::as_tibble(mtcars)
+  mtcars$vs <- as.factor(mtcars$vs)
+
+  rec <- recipes::recipe(mpg ~ ., data = mtcars) %>%
+    themis::step_downsample(vs, skip = TRUE) %>%
+    recipes::prep()
+
+  res <- dplyr::mutate(mtcars, !!!orbital_inline(orbital(rec)))
+
+  exp <- recipes::bake(rec, new_data = mtcars)
+  exp <- exp[names(res)]
+
+  expect_equal(res, exp)
+})
+
+test_that("step_downsample errors with skip = FALSE", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("themis")
+
+  mtcars <- dplyr::as_tibble(mtcars)
+  mtcars$vs <- as.factor(mtcars$vs)
+
+  rec <- recipes::recipe(mpg ~ ., data = mtcars) %>%
+    themis::step_downsample(vs, skip = FALSE) %>%
+    recipes::prep()
+
+  expect_snapshot(
+    error = TRUE,
+    orbital(rec)
+  )
+})
+
+test_that("spark - step_downsample works", {
+  skip_if_not_installed("recipes")
+  skip_if_not_installed("themis")
+  skip_if_not_installed("sparklyr")
+  skip_if(is.na(testthat_spark_env_version()))
+
+  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)))
+  res$vs <- as.character(res$vs)
+
+  sc <- testthat_spark_connection()
+  mtcars_tbl <- testthat_tbl("mtcars_downsample")
+
+  res_spark <- dplyr::mutate(mtcars_tbl, !!!orbital_inline(orbital(rec))) %>%
+    dplyr::collect()
+
+  expect_equal(res_spark, res)
+})