refactor: extract internal tuned values in instance (#453)

* refactor: extract internal tuned values in instance

* ...

* ...

* ...

R/TuningInstanceAsyncMulticrit.R

@@ -118,9 +118,14 @@ TuningInstanceAsyncMultiCrit = R6Class("TuningInstanceAsyncMultiCrit",
     #'
     #' @param ydt (`numeric(1)`)\cr
     #'  Optimal outcomes, e.g. the Pareto front.
-    #' @param ... (`any`)\cr
-    #' ignored.
-    assign_result = function(xdt, ydt, learner_param_vals = NULL, ...) {
+    #' @param xydt (`data.table::data.table()`)\cr
+    #'   Point, outcome, and additional information.
+    assign_result = function(xdt, ydt, learner_param_vals = NULL, xydt = NULL) {
+      # extract internal tuned values
+      if ("internal_tuned_values" %in% names(xydt)) {
+        set(xdt, j = "internal_tuned_values", value = list(xydt[["internal_tuned_values"]]))
+      }
+
       # set the column with the learner param_vals that were not optimized over but set implicitly
       if (is.null(learner_param_vals)) {
         learner_param_vals = self$objective$learner$param_set$values

R/TuningInstanceAsyncSingleCrit.R

@@ -128,12 +128,17 @@ TuningInstanceAsyncSingleCrit = R6Class("TuningInstanceAsyncSingleCrit",
     #'
     #' @param y (`numeric(1)`)\cr
     #' Optimal outcome.
-    #' @param ... (`any`)\cr
-    #' ignored.
-    assign_result = function(xdt, y, learner_param_vals = NULL, ...) {
+    #' @param xydt (`data.table::data.table()`)\cr
+    #'   Point, outcome, and additional information.
+    assign_result = function(xdt, y, learner_param_vals = NULL, xydt = NULL) {
       # set the column with the learner param_vals that were not optimized over but set implicitly
       assert_list(learner_param_vals, null.ok = TRUE, names = "named")
 
+      # extract internal tuned values
+      if ("internal_tuned_values" %in% names(xydt)) {
+        set(xdt, j = "internal_tuned_values", value = list(xydt[["internal_tuned_values"]]))
+      }
+
       if (is.null(learner_param_vals)) {
         learner_param_vals = self$objective$learner$param_set$values
       }

R/TuningInstanceBatchMulticrit.R

@@ -152,9 +152,14 @@ TuningInstanceBatchMultiCrit = R6Class("TuningInstanceBatchMultiCrit",
     #'
     #' @param ydt (`data.table::data.table()`)\cr
     #'   Optimal outcomes, e.g. the Pareto front.
-    #' @param ... (`any`)\cr
-    #' ignored.
-    assign_result = function(xdt, ydt, learner_param_vals = NULL) {
+    #' @param xydt (`data.table::data.table()`)\cr
+    #'   Point, outcome, and additional information.
+    assign_result = function(xdt, ydt, learner_param_vals = NULL, xydt = NULL) {
+      # extract internal tuned values
+      if ("internal_tuned_values" %in% names(xydt)) {
+        set(xdt, j = "internal_tuned_values", value = list(xydt[["internal_tuned_values"]]))
+      }
+
       # set the column with the learner param_vals that were not optimized over but set implicitly
       if (is.null(learner_param_vals)) {
         learner_param_vals = self$objective$learner$param_set$values

R/TuningInstanceBatchSingleCrit.R

@@ -191,12 +191,18 @@ TuningInstanceBatchSingleCrit = R6Class("TuningInstanceBatchSingleCrit",
     #'
     #' @param y (`numeric(1)`)\cr
     #'   Optimal outcome.
-    #' @param ... (`any`)\cr
-    #' ignored.
-    assign_result = function(xdt, y, learner_param_vals = NULL, ...) {
+    #' @param xydt (`data.table::data.table()`)\cr
+    #'   Point, outcome, and additional information.
+    assign_result = function(xdt, y, learner_param_vals = NULL, xydt = NULL) {
+
       # set the column with the learner param_vals that were not optimized over but set implicitly
       assert_list(learner_param_vals, null.ok = TRUE, names = "named")
 
+      # extract internal tuned values
+      if ("internal_tuned_values" %in% names(xydt)) {
+        set(xdt, j = "internal_tuned_values", value = list(xydt[["internal_tuned_values"]]))
+      }
+
       # learner param values
       if (is.null(learner_param_vals)) {
         learner_param_vals = self$objective$learner$param_set$values

tests/testthat/test_Tuner.R

@@ -51,6 +51,8 @@ test_that("we get a result when some subordinate params are not fulfilled", {
 })
 
 test_that("print method workds", {
+  skip_if_not_installed("GenSA")
+
   param_set = ps(p1 = p_lgl())
   param_set$values$p1 = TRUE
   param_classes = "ParamLgl"
@@ -125,6 +127,8 @@ test_that("Tuner works with instantiated resampling", {
 })
 
 test_that("Tuner active bindings work", {
+  skip_if_not_installed("GenSA")
+
   param_set = ps(p1 = p_lgl())
   param_set$values$p1 = TRUE
   param_classes = "ParamLgl"

tests/testthat/test_TunerBatchCmaes.R

@@ -1,6 +1,7 @@
 test_that("TunerBatchCmaes", {
+  skip_if_not_installed("adagio")
 
-expect_tuner(tnr("cmaes"))
+  expect_tuner(tnr("cmaes"))
 
   learner = lrn("classif.rpart",
     cp = to_tune(1e-04, 1e-1, logscale = TRUE),

tests/testthat/test_TunerBatchNLoptr.R

@@ -1,4 +1,6 @@
 test_that("TunerNLoptr", {
   skip_on_os("windows")
+  skip_if_not_installed("nloptr")
+
   test_tuner("nloptr", algorithm = "NLOPT_LN_BOBYQA", term_evals = 4)
 })

tests/testthat/test_mlr_tuners.R

@@ -1,4 +1,6 @@
 test_that("mlr_tuners", {
+  skip_if_not_installed(c("rush", "adagio", "GenSA", "irace", "nloptr"))
+
   expect_dictionary(mlr_tuners, min_items = 1L)
   keys = mlr_tuners$keys()
 
@@ -14,6 +16,8 @@ test_that("mlr_tuners sugar", {
 })
 
 test_that("as.data.table objects parameter", {
+  skip_if_not_installed(c("rush", "adagio", "GenSA", "irace", "nloptr"))
+
   tab = as.data.table(mlr_tuners, objects = TRUE)
   expect_data_table(tab)
   expect_list(tab$object, "Tuner", any.missing = FALSE)