refactor: move to mlr3learners style autotest, pass params in lrn call directly (#62)

* refactor: move to mlr3learners style autotest, pass params in lrn call directyl

* refactor: use explicit integers

Author: m-muecke

R/LearnerClust.R

@@ -28,8 +28,8 @@
 #' ids
 #'
 #' # get a specific learner from mlr_learners:
-#' lrn = mlr_learners$get("clust.kmeans")
-#' print(lrn)
+#' learner = lrn("clust.kmeans")
+#' print(learner)
 LearnerClust = R6Class("LearnerClust",
   inherit = Learner,
   public = list(

README.md

@@ -77,12 +77,12 @@ Also, the package is integrated with **[mlr3viz](https://github.com/mlr-org/mlr3
 
 ## Example
 
-```{r}
+```r
 library(mlr3)
 library(mlr3cluster)
 
-task = mlr_tasks$get("usarrests")
-learner = mlr_learners$get("clust.kmeans")
+task = tsk("usarrests")
+learner = lrn("clust.kmeans")
 learner$train(task)
 preds = learner$predict(task = task)
 ```

man/LearnerClust.Rd

@@ -1,7 +1,6 @@
 test_that("predict on newdata works / clust", {
   task = tsk("usarrests")$filter(1:40)
-  learner = lrn("clust.featureless")
-  learner$param_set$values = list(num_clusters = 1L)
+  learner = lrn("clust.featureless", num_clusters = 1L)
   expect_error(learner$predict(task), "trained")
   learner$train(task)
   expect_task(learner$state$train_task)
@@ -15,26 +14,24 @@ test_that("predict on newdata works / clust", {
 
   # rely on internally stored task representation
   p = learner$predict_newdata(newdata = newdata, task = NULL)
-  expect_data_table(as.data.table(p), nrows = 10)
+  expect_data_table(as.data.table(p), nrows = 10L)
   expect_set_equal(as.data.table(p)$row_ids, 1:10)
   expect_null(p$truth)
 })
 
 test_that("reset()", {
   task = tsk("usarrests")
-  lrn = lrn("clust.featureless")
-  lrn$param_set$values = list(num_clusters = 2L)
+  learner = lrn("clust.featureless", num_clusters = 2L)
 
-  lrn$train(task)
-  expect_list(lrn$state, names = "unique")
-  expect_learner(lrn$reset())
-  expect_null(lrn$state)
+  learner$train(task)
+  expect_list(learner$state, names = "unique")
+  expect_learner(learner$reset())
+  expect_null(learner$state)
 })
 
 test_that("empty predict set (#421)", {
   task = tsk("usarrests")
-  learner = lrn("clust.featureless")
-  learner$param_set$values = list(num_clusters = 1L)
+  learner = lrn("clust.featureless", num_clusters = 1L)
   resampling = rsmp("holdout", ratio = 1)
   hout = resampling$instantiate(task)
   model = learner$train(task, hout$train_set(1))

tests/testthat/test_LearnerClust.R

@@ -1,10 +1,8 @@
 test_that("Cluster measures", {
   keys = mlr_measures$keys("clust")
   task = tsk("usarrests")
-  learner = mlr_learners$get("clust.kmeans")
-  learner$param_set$values = list(centers = 2)
-  learner$train(task)
-  p = learner$predict(task)
+  learner = lrn("clust.kmeans", centers = 2L)
+  p = learner$train(task)$predict(task)
 
   for (key in keys) {
     m = mlr_measures$get(key)

tests/testthat/test_MeasureClust.R

@@ -9,18 +9,16 @@ test_that("Construction", {
 
 test_that("Internally constructed Prediction", {
   task = tsk("usarrests")
-  lrn = mlr_learners$get("clust.featureless")
-  lrn$param_set$values = list(num_clusters = 1L)
-  p = lrn$train(task)$predict(task)
+  learner = lrn("clust.featureless", num_clusters = 1L)
+  p = learner$train(task)$predict(task)
   expect_prediction(p)
   expect_prediction_clust(p)
 })
 
 test_that("filter works", {
   task = tsk("usarrests")
-  lrn = mlr_learners$get("clust.featureless")
-  lrn$param_set$values = list(num_clusters = 1L)
-  p = lrn$train(task)$predict(task)
+  learner = lrn("clust.featureless", num_clusters = 1L)
+  p = learner$train(task)$predict(task)
   pdata = p$data
 
   pdata = filter_prediction_data(pdata, row_ids = 1:3)

tests/testthat/test_PredictionClust.R

@@ -2,14 +2,14 @@ test_that("Basic ops on usarrests task", {
   task = tsk("usarrests")
   expect_task(task)
   expect_task_clust(task)
-  expect_identical(task$target_names, character(0))
+  expect_identical(task$target_names, character(0L))
 })
 
 test_that("Basic ops on ruspini task", {
   task = tsk("ruspini")
   expect_task(task)
   expect_task_clust(task)
-  expect_identical(task$target_names, character(0))
+  expect_identical(task$target_names, character(0L))
 })
 
 test_that("0 feature task", {
@@ -22,8 +22,7 @@ test_that("0 feature task", {
   expect_task_clust(task)
   expect_data_table(task$data(), ncols = 1L)
 
-  lrn = lrn("clust.featureless")
-  lrn$param_set$values = list(num_clusters = 3L)
-  p = lrn$train(task)$predict(task)
+  learner = lrn("clust.featureless", num_clusters = 3L)
+  p = learner$train(task)$predict(task)
   expect_prediction(p)
 })

tests/testthat/test_TaskClust.R

@@ -3,15 +3,13 @@ skip_if_not_installed("clue")
 test_that("autotest", {
   learner = mlr3::lrn("clust.agnes")
   expect_learner(learner)
-
   result = run_autotest(learner)
   expect_true(result, info = result$error)
 })
 
-
 test_that("Learner properties are respected", {
   task = tsk("usarrests")
-  learner = mlr_learners$get("clust.agnes")
+  learner = lrn("clust.agnes")
   expect_learner(learner, task)
 
   # test on multiple paramsets

tests/testthat/test_mlr_learners_clust_agnes.R

@@ -1,18 +1,15 @@
 skip_if_not_installed("apcluster")
 
 test_that("autotest", {
-  learner = mlr3::lrn("clust.ap")
-  learner$param_set$values = list(s = apcluster::negDistMat(r = 2L))
+  learner = mlr3::lrn("clust.ap", s = apcluster::negDistMat(r = 2L))
   expect_learner(learner)
-
   result = run_autotest(learner)
   expect_true(result, info = result$error)
 })
 
-
 test_that("Learner properties are respected", {
   task = tsk("usarrests")
-  learner = mlr_learners$get("clust.ap")
+  learner = lrn("clust.ap")
   expect_learner(learner, task)
 
   # test on multiple paramsets

tests/testthat/test_mlr_learners_clust_ap.R

@@ -3,19 +3,17 @@ skip_if_not_installed("e1071")
 test_that("autotest", {
   learner = mlr3::lrn("clust.cmeans")
   expect_learner(learner)
-
   result = run_autotest(learner)
   expect_true(result, info = result$error)
 })
 
-
 test_that("Learner properties are respected", {
   task = tsk("usarrests")
-  learner = mlr_learners$get("clust.cmeans")
+  learner = lrn("clust.cmeans")
   expect_learner(learner, task)
 
   # test on multiple paramsets
-  centers = data.frame(matrix(ncol = length(colnames(task$data())), nrow = 4))
+  centers = data.frame(matrix(ncol = length(colnames(task$data())), nrow = 4L))
   colnames(centers) = colnames(task$data())
   centers$Assault = c(100, 200, 150, 300)
   centers$Murder = c(11, 3, 10, 5)