predict_newdata_fast as thin wrapper around predict_newdata

NEWS.md

@@ -1,5 +1,7 @@
 # mlr3pipelines 0.10.0-9000
 
+* New method `$predict_newdata_fast()` for `GraphLearner`. Note that currently this is only a thin wrapper around `$predict_newdata()` to maintain compatibility, but in the future it may get optimized to enable faster predictions on new data.
+
 # mlr3pipelines 0.10.0
 
 * Pretty-printing some info using the `cli` package now.

R/GraphLearner.R

@@ -118,7 +118,16 @@
 #'   corresponding [`PipeOpBranch`] is searched, and its hyperparameter configuration is used to select the base learner.
 #'   There may be multiple corresponding [`PipeOpBranch`]s, which are all considered.
 #'   If `resolve_branching` is `FALSE`, [`PipeOpUnbranch`] is treated as any other `PipeOp` with multiple inputs; all possible branch paths are considered equally.
-#'
+#' * `predict_newdata_fast(newdata, task = NULL)`\cr
+#'   (`data.frame`, [`Task`][mlr3::Task] | `NULL`) -> [`Prediction`][mlr3::Prediction]\cr
+#'   Predicts outcomes for new data in `newdata` using the model fitted during `$train()`.\cr
+#'   For the moment, this is merely a thin wrapper around [`Learner$predict_newdata()`][mlr3::Learner] to ensure compatibility, meaning that *no speedup* is currently achieved.
+#'   In the future, this method may be optimized to be faster than `$predict_newdata()`.\cr
+#'   Unlike `$predict_newdata()`, this method does not return a [Prediction] object.
+#'   Instead, it returns a list with elements depending on `$task_type` and `$predict_type`:
+#'   * for `task_type = "classif"`: `response` and `prob`, or `quantiles` (if `predict_type = "quantiles"`)
+#'   * for `task_type = "regr"`: `response` and `se`
+#' 
 #' The following standard extractors as defined by the [`Learner`][mlr3::Learner] class are available.
 #' Note that these typically only extract information from the `$base_learner()`.
 #' This works well for simple [`Graph`]s that do not modify features too much, but may give unexpected results for `Graph`s that
@@ -324,6 +333,18 @@ GraphLearner = R6Class("GraphLearner", inherit = Learner,
         pp = "non-sequential"
       }
       cat_cli(cli_h3("Pipeline: {.strong {pp}}"))
+    },
+    # TODO: Optimize this method to be actually faster than predict_newdata(), #968
+    predict_newdata_fast = function(newdata, task = NULL) {
+      pred = self$predict_newdata(newdata, task)
+      if (self$task_type == "regr") {
+        if (!is.null(pred$quantiles)) {
+          return(list(quantiles = pred$quantiles))
+        }
+        list(response = pred$response, se = if (!all(is.na(pred$se))) pred$se else NULL)
+      } else if (self$task_type == "classif") {
+        list(response = pred$response, prob = pred$prob)
+      }
     }
   ),
   active = list(

tests/testthat/test_GraphLearner.R

@@ -1345,3 +1345,52 @@ test_that("GraphLearner other properties", {
   expect_true(all(c("loglik", "oob_error") %in% g_nested$properties))
 
 })
+
+test_that("GraphLearner - predict_newdata_fast", {
+  # Classification
+  task = tsk("iris")
+  newdata = task$data(cols = task$feature_names, rows = 1:10)
+
+  ## predict_type = "response"
+  learner = lrn("classif.featureless")
+  set.seed(20251117)
+  lrn_pred = learner$train(task)$predict_newdata_fast(newdata)
+  set.seed(20251117)
+  glrn_pred = as_learner(po("learner", learner))$train(task)$predict_newdata_fast(newdata)
+
+  expect_equal(as.character(glrn_pred$response), lrn_pred$response)
+  expect_equal(glrn_pred$prob, lrn_pred$prob)
+
+  ## predict_type = "prob"
+  learner = lrn("classif.featureless", predict_type = "prob")
+  set.seed(20251117)
+  lrn_pred = learner$train(task)$predict_newdata_fast(newdata)
+  set.seed(20251117)
+  glrn_pred = as_learner(po("learner", learner))$train(task)$predict_newdata_fast(newdata)
+
+  # Some Learners only return probs if predict_type is "prob" in .predict() and labels are inferred later, 
+  # so there is an acceptable difference between the two methods. 
+  expect_true(identical(glrn_pred$response, lrn_pred$response) || is.null(lrn_pred$response))
+  expect_equal(glrn_pred$prob, lrn_pred$prob)
+
+  # Regression
+  task = tsk("mtcars")
+  newdata = task$data(cols = task$feature_names, rows = 1:10)
+  ## predict_type = "response"
+  learner = lrn("regr.featureless")
+  lrn_pred = learner$train(task)$predict_newdata_fast(newdata)
+  glrn_pred = as_learner(po("learner", learner))$train(task)$predict_newdata_fast(newdata)
+  expect_equal(glrn_pred, lrn_pred)
+
+  ## predict_type = "se"
+  learner = lrn("regr.featureless", predict_type = "se")
+  lrn_pred = learner$train(task)$predict_newdata_fast(newdata)
+  glrn_pred = as_learner(po("learner", learner))$train(task)$predict_newdata_fast(newdata)
+  expect_equal(glrn_pred, lrn_pred)
+
+  ## predict_type = "quantiles"
+  learner = lrn("regr.featureless", predict_type = "quantiles", quantiles = 0.25)
+  lrn_pred = learner$train(task)$predict_newdata_fast(newdata)
+  glrn_pred = as_learner(po("learner", learner))$train(task)$predict_newdata_fast(newdata)
+  expect_equal(glrn_pred$quantiles[, 1L], lrn_pred$quantiles[, 1L])
+})