refactor: use tsk(), stopf(), and minor lints

.lintr

@@ -5,6 +5,5 @@ linters: linters_with_defaults(
     object_name_linter = object_name_linter(c("snake_case", "CamelCase")), # only allow snake case and camel case object names
     cyclocomp_linter = NULL, # do not check function complexity
     commented_code_linter = NULL, # allow code in comments
-    line_length_linter = line_length_linter(120)
-    )
-
+    line_length_linter = line_length_linter(120L)
+    )

R/LearnerClassifXgboost.R

@@ -220,7 +220,7 @@ LearnerClassifXgboost = R6Class("LearnerClassifXgboost",
       }
 
       if (self$predict_type == "prob" && pv$objective == "multi:softmax") {
-        stop("objective = 'multi:softmax' does not work with predict_type = 'prob'")
+        stopf("objective = 'multi:softmax' does not work with predict_type = 'prob'")
       }
 
       switch(pv$objective,
@@ -311,7 +311,7 @@ LearnerClassifXgboost = R6Class("LearnerClassifXgboost",
       pars = self$param_set$get_values(tags = "train")
       pars_train = self$state$param_vals
       if (!is.null(pars_train$early_stopping_rounds)) {
-        stop("The parameter `early_stopping_rounds` is set. Early stopping and hotstarting are incompatible.")
+        stopf("The parameter `early_stopping_rounds` is set. Early stopping and hotstarting are incompatible.")
       }
 
       # Calculate additional boosting iterations

R/LearnerRegrKM.R

@@ -81,10 +81,8 @@ LearnerRegrKM = R6Class("LearnerRegrKM",
       data = as_numeric_matrix(task$data(cols = task$feature_names))
       truth = task$truth()
 
-      if (!is.null(pv$optim.method)) {
-        if (pv$optim.method == "gen" && !requireNamespace("rgenoud", quietly = TRUE)) {
-          stop("The 'rgenoud' package is required for optimization method 'gen'.")
-        }
+      if (!is.null(pv$optim.method) && pv$optim.method == "gen" && !requireNamespace("rgenoud", quietly = TRUE)) {
+        stopf("The 'rgenoud' package is required for optimization method 'gen'.")
       }
 
       ns = pv$nugget.stability

R/LearnerRegrNnet.R

@@ -58,9 +58,9 @@ LearnerRegrNnet = R6Class("LearnerRegrNnet",
         id = "regr.nnet",
         packages = c("mlr3learners", "nnet"),
         feature_types = c("numeric", "factor", "ordered", "integer"),
-        predict_types = c("response"),
+        predict_types = "response",
         param_set = ps,
-        properties = c("weights"),
+        properties = "weights",
         label = "Single Layer Neural Network",
         man = "mlr3learners::mlr_learners_regr.nnet"
       )

R/LearnerRegrXgboost.R

@@ -234,7 +234,7 @@ LearnerRegrXgboost = R6Class("LearnerRegrXgboost",
       pars = self$param_set$get_values(tags = "train")
       pars_train = self$state$param_vals
       if (!is.null(pars_train$early_stopping_rounds)) {
-        stop("The parameter `early_stopping_rounds` is set. Early stopping and hotstarting are incompatible.")
+        stopf("The parameter `early_stopping_rounds` is set. Early stopping and hotstarting are incompatible.")
       }
 
       # Calculate additional boosting iterations

inst/paramtest/test_paramtest_classif.xgboost.R

@@ -8,19 +8,19 @@ xp = rvest::html_elements(x, "p")
 x = c(rvest::html_text2(xli), rvest::html_text2(xp))
 
 add_params_xgboost = x %>%
-  grep("default=", ., value = T) %>%
-  strsplit(., split = " ") %>%
-  mlr3misc::map_chr(., function(x) x[1]) %>%
-  gsub(",", replacement = "", .) %>%
+  grep("default=", ., value = TRUE) %>%
+  strsplit(., split = " ", fixed = TRUE) %>%
+  mlr3misc::map_chr(1L) %>%
+  gsub(",", replacement = "", ., fixed = TRUE) %>%
   ## these are defined on the same line as colsample_bytree and cannot be scraped therefore
   append(values = c("colsample_bylevel", "colsample_bynode")) %>%
   # values which do not match regex
   append(values = c("interaction_constraints", "monotone_constraints", "base_score")) %>%
   # only defined in help page but not in signature or website
-  append(values = c("lambda_bias"))
+  append(values = "lambda_bias")
 
 test_that("classif.xgboost", {
-  learner = lrn("classif.xgboost", nrounds = 1)
+  learner = lrn("classif.xgboost", nrounds = 1L)
   fun = list(xgboost::xgb.train, xgboost::xgboost, add_params_xgboost)
   exclude = c(
     "x", # handled by mlr3

inst/paramtest/test_paramtest_regr.xgboost.R

@@ -8,19 +8,19 @@ xp = rvest::html_elements(x, "p")
 x = c(rvest::html_text2(xli), rvest::html_text2(xp))
 
 add_params_xgboost = x %>%
-  grep("default=", ., value = T) %>%
-  strsplit(., split = " ") %>%
-  mlr3misc::map_chr(., function(x) x[1]) %>%
-  gsub(",", replacement = "", .) %>%
+  grep("default=", ., value = TRUE) %>%
+  strsplit(., split = " ", fixed = TRUE) %>%
+  mlr3misc::map_chr(1L) %>%
+  gsub(",", replacement = "", ., fixed = TRUE) %>%
   ## these are defined on the same line as colsample_bytree and cannot be scraped therefore
   append(values = c("colsample_bylevel", "colsample_bynode")) %>%
   # values which do not match regex
   append(values = c("interaction_constraints", "monotone_constraints", "base_score")) %>%
   # only defined in help page but not in signature or website
-  append(values = c("lambda_bias"))
+  append(values = "lambda_bias")
 
 test_that("regr.xgboost", {
-  learner = lrn("regr.xgboost", nrounds = 1)
+  learner = lrn("regr.xgboost", nrounds = 1L)
   fun = list(xgboost::xgb.train, xgboost::xgboost, add_params_xgboost)
   exclude = c(
     "x", # handled by mlr3

tests/testthat/test_classif_glmnet.R

@@ -2,7 +2,7 @@ skip_on_os("solaris") # glmnet not working properly on solaris
 skip_if_not_installed("glmnet")
 
 test_that("autotest", {
-  learner = mlr3::lrn("classif.glmnet", lambda = .1)
+  learner = mlr3::lrn("classif.glmnet", lambda = 0.1)
   expect_learner(learner)
 
   skip_on_os("solaris")
@@ -12,7 +12,7 @@ test_that("autotest", {
 
 test_that("prob column reordering (#155)", {
   task = tsk("sonar")
-  learner = mlr3::lrn("classif.glmnet", predict_type = "prob", lambda = .1)
+  learner = mlr3::lrn("classif.glmnet", predict_type = "prob", lambda = 0.1)
 
   task$positive = "M"
   learner$train(task)

tests/testthat/test_classif_ranger.R

@@ -56,7 +56,7 @@ test_that("mtry.ratio", {
 
 test_that("convert_ratio", {
   task = tsk("sonar")
-  learner = lrn("classif.ranger", num.trees = 5, mtry.ratio = .5)
+  learner = lrn("classif.ranger", num.trees = 5, mtry.ratio = 0.5)
   expect_equal(learner$train(task)$model$mtry, 30)
 
   learner$param_set$values$mtry.ratio = 0

tests/testthat/test_classif_xgboost.R

@@ -15,14 +15,14 @@ test_that("xgboost with softmax", {
 })
 
 test_that("xgboost with multi:softprob", {
-  task = mlr_tasks$get("sonar")
+  task = tsk("sonar")
   learner = mlr3::lrn("classif.xgboost", nrounds = 5L, objective = "multi:softprob")
   p = learner$train(task)$predict(task)
   expect_equal(unname(p$score()), 0)
 })
 
 test_that("xgboost with binary:logistic", {
-  task = mlr_tasks$get("sonar")
+  task = tsk("sonar")
   learner = mlr3::lrn("classif.xgboost", nrounds = 5L)
   p = learner$train(task)$predict(task)
   expect_equal(unname(p$score()), 0)

tests/testthat/test_regr_glmnet.R

@@ -2,7 +2,7 @@ skip_on_os("solaris") # glmnet not working properly on solaris
 skip_if_not_installed("glmnet")
 
 test_that("autotest", {
-  learner = mlr3::lrn("regr.glmnet", lambda = .1)
+  learner = mlr3::lrn("regr.glmnet", lambda = 0.1)
   expect_learner(learner)
 
   skip_on_os("solaris")