Merge pull request #765 from mlr-org/rename_fu_in_ppl

rename featureunion in ppls
mlr-org · Mar 26, 2024 · 7d18533 · 7d18533
2 parents 5dc84f6 + 47e3ee5
commit 7d18533
Show file tree

Hide file tree

Showing 5 changed files with 22 additions and 6 deletions.
diff --git a/NEWS.md b/NEWS.md
@@ -1,5 +1,6 @@
 # mlr3pipelines 0.5.0-9000
 
+* Changed the ID of `PipeOpFeatureUnion` used in `ppl("robustify")` and `ppl("stacking")`.
 * `pipeline_bagging()` gets the `replace` argument (old behaviour `FALSE` by default).
 * Feature: The `$add_pipeop()` method got an argument `clone` (old behaviour `TRUE` by default).
 * Bugfix: `PipeOpFeatureUnion` in some rare cases dropped variables called `"x"`.

diff --git a/R/pipeline_robustify.R b/R/pipeline_robustify.R
@@ -170,7 +170,7 @@ pipeline_robustify = function(task = NULL, learner = NULL,
       imputing,
       po("missind", affect_columns = selector_type(c("numeric", "integer", "logical")), type = if (missind_numeric) "numeric" else "factor")
     )),
-    if (has_numbers || has_logicals) po("featureunion"),
+    if (has_numbers || has_logicals) po("featureunion", id = "featureunion_robustify"),
     if (has_factorials) po("imputeoor")
   )
 

diff --git a/R/pipeline_stacking.R b/R/pipeline_stacking.R
@@ -53,7 +53,7 @@ pipeline_stacking = function(base_learners, super_learner, method = "cv", folds
   if (use_features) base_learners_cv = c(base_learners_cv, po("nop"))
 
   gunion(base_learners_cv, in_place = TRUE) %>>!%
-     po("featureunion") %>>!%
+     po("featureunion", id = "featureunion_stacking") %>>!%
      super_learner
 }
 

diff --git a/tests/testthat/test_mlr_graphs_stacking.R b/tests/testthat/test_mlr_graphs_stacking.R
@@ -10,7 +10,7 @@ test_that("Stacking Pipeline", {
   # default
   graph_stack = pipeline_stacking(base_learners, super_learner)
   expect_graph(graph_stack)
-  expect_names(graph_stack$ids(), identical.to = c("base.rpart", "nop", "featureunion", "super.rpart"))
+  expect_names(graph_stack$ids(), identical.to = c("base.rpart", "nop", "featureunion_stacking", "super.rpart"))
   graph_learner = as_learner(graph_stack)
   graph_learner$train(tsk("iris"))
   expect_class(graph_learner$model$super.rpart$model, "rpart")
@@ -19,7 +19,7 @@ test_that("Stacking Pipeline", {
   # no nop
   graph_stack = pipeline_stacking(base_learners, super_learner, use_features = FALSE)
   expect_graph(graph_stack)
-  expect_names(graph_stack$ids(), identical.to = c("base.rpart", "featureunion", "super.rpart"))
+  expect_names(graph_stack$ids(), identical.to = c("base.rpart", "featureunion_stacking", "super.rpart"))
   graph_learner = as_learner(graph_stack)
   graph_learner$train(tsk("iris"))
   expect_class(graph_learner$model$super.rpart$model, "rpart")
@@ -28,7 +28,7 @@ test_that("Stacking Pipeline", {
   # folds
   graph_stack = pipeline_stacking(base_learners, super_learner, folds = 5)
   expect_graph(graph_stack)
-  expect_names(graph_stack$ids(), identical.to = c("base.rpart", "nop", "featureunion", "super.rpart"))
+  expect_names(graph_stack$ids(), identical.to = c("base.rpart", "nop", "featureunion_stacking", "super.rpart"))
   graph_learner = as_learner(graph_stack)
   graph_learner$train(tsk("iris"))
   expect_equal(graph_learner$graph$pipeops$base.rpart$param_set$values$resampling.folds, 5)
@@ -38,7 +38,7 @@ test_that("Stacking Pipeline", {
   # insample
   graph_stack = pipeline_stacking(base_learners, super_learner, method = "insample")
   expect_graph(graph_stack)
-  expect_names(graph_stack$ids(), identical.to = c("base.rpart", "nop", "featureunion", "super.rpart"))
+  expect_names(graph_stack$ids(), identical.to = c("base.rpart", "nop", "featureunion_stacking", "super.rpart"))
   graph_learner = as_learner(graph_stack)
   graph_learner$train(tsk("iris"))
   expect_equal(graph_learner$graph$pipeops$base.rpart$param_set$values$resampling.method, "insample")

diff --git a/tests/testthat/test_ppl.R b/tests/testthat/test_ppl.R
@@ -57,3 +57,18 @@ test_that("mlr_pipeops multi-access works", {
   expect_equal(ppls(), mlr_graphs)
 
 })
+
+test_that("mlr3book authors don't sleepwalk through life", {
+
+  tasks = tsks(c("breast_cancer", "sonar"))
+
+  glrn_stack = as_learner(ppl("robustify") %>>% ppl("stacking",
+      lrns(c("classif.rpart", "classif.debug")),
+      lrn("classif.rpart", id = "classif.rpart2")
+  ))
+  glrn_stack$id = "Stack"
+
+  learners = c(glrn_stack)
+  bmr = benchmark(benchmark_grid(tasks, learners, rsmp("cv", folds = 2)))
+
+})