mlr-org · pat-s · Aug 23, 2021 · Aug 5, 2021 · Aug 5, 2021
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -76,6 +76,7 @@ Collate:
     'FilterPerformance.R'
     'FilterPermutation.R'
     'FilterRelief.R'
+    'FilterSelectedFeatures.R'
     'FilterVariance.R'
     'flt.R'
     'helper.R'

diff --git a/NAMESPACE b/NAMESPACE
@@ -21,6 +21,7 @@ export(FilterNJMIM)
 export(FilterPerformance)
 export(FilterPermutation)
 export(FilterRelief)
+export(FilterSelectedFeatures)
 export(FilterVariance)
 export(as.data.table)
 export(flt)

diff --git a/R/FilterSelectedFeatures.R b/R/FilterSelectedFeatures.R
@@ -0,0 +1,65 @@
+#' @title Filter for Embedded Feature Selection
+#'
+#' @name mlr_filters_selected_features
+#'
+#' @description
+#' Filter using embedded feature selection of machine learning algorithms.
+#' Takes a [mlr3::Learner] which is capable of extracting the selected features
+#' (property "selected_features"), fits the model and extracts the selected
+#' features.
+#'
+#' Note that contrary to [mlr_filters_importance], there is no ordering in
+#' the selected features. Selected features get a score of 1, deselected
+#' features get a score of 0. The order of selected features is random and
+#' different from the order in the learner. In combination with
+#' \CRANpkg{mlr3pipelines}, only the filter criterion `cutoff` makes sense.
+#'
+#'
+#' @family Filter
+#' @template seealso_filter
+#' @export
+#' @examples
+#' task = mlr3::tsk("iris")
+#' learner = mlr3::lrn("classif.rpart")
+#' filter = flt("selected_features", learner = learner)
+#' filter$calculate(task)
+#' as.data.table(filter)
+FilterSelectedFeatures = R6Class("FilterSelectedFeatures",
+  inherit = Filter,
+
+  public = list(
+
+    #' @field learner ([mlr3::Learner])\cr
+    #'   Learner to extract the importance values from.
+    learner = NULL,
+
+    #' @description Create a FilterImportance object.
+    #' @param learner ([mlr3::Learner])\cr
+    #'   Learner to extract the importance values from.
+    initialize = function(learner = mlr3::lrn("classif.rpart")) {
+      self$learner = learner = assert_learner(as_learner(learner, clone = TRUE),
+        properties = "selected_features")
+
+      super$initialize(
+        id = "selected_features",
+        task_type = learner$task_type,
+        feature_types = learner$feature_types,
+        packages = learner$packages,
+        param_set = learner$param_set,
+        man = "mlr3filters::mlr_filters_selected_features"
+      )
+    }
+  ),
+
+  private = list(
+    .calculate = function(task, nfeat) {
+      learner = self$learner$clone(deep = TRUE)
+      learner = learner$train(task = task)
+      score = named_vector(task$feature_names, init = 0)
+      replace(score, names(score) %in% learner$selected_features(), 1)
+    }
+  )
+)
+
+#' @include mlr_filters.R
+mlr_filters$add("selected_features", FilterSelectedFeatures)
diff --git a/README.Rmd b/README.Rmd
@@ -7,6 +7,9 @@ output: github_document
 Package website: [release](https://mlr3filters.mlr-org.com/) | [dev](https://mlr3filters.mlr-org.com/dev/)
 
 {mlr3filters} adds feature selection filters to [mlr3](https://mlr3.mlr-org.com).
+The implemented filters can be used stand-alone, or as part of a machine learning pipeline in combination with
+[mlr3pipelines](https://mlr3pipelines.mlr-org.com) and the [filter operator](https://mlr3pipelines.mlr-org.com/reference/mlr_pipeops_filter.html).
+
 Wrapper methods for feature selection are implemented in [mlr3fselect](https://mlr3fselect.mlr-org.com).
 Learners which support the extraction feature importance scores can be combined with a filter from this package for embedded feature selection.
 

diff --git a/man/Filter.Rd b/man/Filter.Rd
diff --git a/man/mlr_filters.Rd b/man/mlr_filters.Rd
diff --git a/man/mlr_filters_anova.Rd b/man/mlr_filters_anova.Rd
diff --git a/man/mlr_filters_auc.Rd b/man/mlr_filters_auc.Rd
diff --git a/man/mlr_filters_carscore.Rd b/man/mlr_filters_carscore.Rd
diff --git a/man/mlr_filters_cmim.Rd b/man/mlr_filters_cmim.Rd
diff --git a/man/mlr_filters_correlation.Rd b/man/mlr_filters_correlation.Rd
diff --git a/man/mlr_filters_disr.Rd b/man/mlr_filters_disr.Rd
diff --git a/man/mlr_filters_find_correlation.Rd b/man/mlr_filters_find_correlation.Rd
diff --git a/man/mlr_filters_importance.Rd b/man/mlr_filters_importance.Rd
diff --git a/man/mlr_filters_information_gain.Rd b/man/mlr_filters_information_gain.Rd
diff --git a/man/mlr_filters_jmi.Rd b/man/mlr_filters_jmi.Rd
diff --git a/man/mlr_filters_jmim.Rd b/man/mlr_filters_jmim.Rd
diff --git a/man/mlr_filters_kruskal_test.Rd b/man/mlr_filters_kruskal_test.Rd
diff --git a/man/mlr_filters_mim.Rd b/man/mlr_filters_mim.Rd
diff --git a/man/mlr_filters_mrmr.Rd b/man/mlr_filters_mrmr.Rd
diff --git a/man/mlr_filters_njmim.Rd b/man/mlr_filters_njmim.Rd
diff --git a/man/mlr_filters_performance.Rd b/man/mlr_filters_performance.Rd
diff --git a/man/mlr_filters_permutation.Rd b/man/mlr_filters_permutation.Rd
diff --git a/man/mlr_filters_relief.Rd b/man/mlr_filters_relief.Rd
diff --git a/man/mlr_filters_selected_features.Rd b/man/mlr_filters_selected_features.Rd
diff --git a/man/mlr_filters_variance.Rd b/man/mlr_filters_variance.Rd
diff --git a/tests/testthat/test_FilterSelectedFeatures.R b/tests/testthat/test_FilterSelectedFeatures.R
@@ -0,0 +1,8 @@
+test_that("FilterSelectedFeatures", {
+  set.seed(42)
+  task = mlr3::mlr_tasks$get("wine")
+  learner = mlr3::mlr_learners$get("classif.rpart")
+  f = FilterSelectedFeatures$new(learner = learner)
+  f$calculate(task)
+  expect_filter(f, task = task)
+})