remove Metrics

Author: mllg

DESCRIPTION

@@ -39,6 +39,7 @@ Suggests:
     care,
     FSelectorRcpp,
     lgr,
+    mlr3measures,
     mlr3learners,
     praznik,
     rpart,

R/FilterAUC.R

@@ -12,9 +12,11 @@
 #' flt("auc")
 #' ```
 #'
-#' @description Area under the (ROC) Curve filter calling [Metrics::auc()] from
-#' package \CRANpkg{Metrics}. Returns the absolute value of the difference
-#' between the AUC and 0.5.
+#' @description
+#' Area under the (ROC) Curve filter, analogously to [mlr3measures::auc()] from \CRANpkg{mlr3meaures}.
+#' Missing values of the features are removed before calculating the AUC.
+#' If the AUC is undefined for the input, it is set to 0.5 (random classifier).
+#' The absolute value of the difference between the AUC and 0.5 is used as final filter value.
 #'
 #' @family Filter
 #' @template seealso_filter
@@ -29,21 +31,34 @@ FilterAUC = R6Class("FilterAUC", inherit = Filter,
     initialize = function() {
       super$initialize(
         id = "auc",
-        packages = "Metrics",
+        packages = "mlr3measures",
         feature_types = c("integer", "numeric"),
         task_type = "classif",
         task_properties = "twoclass"
       )
     },
 
     calculate_internal = function(task, nfeat) {
-      x = task$truth() == task$positive
-      y = task$data(cols = task$feature_names)
-      score = map_dbl(y, function(y) Metrics::auc(x, y))
+      y = task$truth() == task$positive
+      x = task$data(cols = task$feature_names)
+      score = map_dbl(x, function(x) {
+        keep = !is.na(x)
+        auc(y[keep], x[keep])
+      })
       abs(0.5 - score)
     }
   )
 )
 
 #' @include mlr_filters.R
 mlr_filters$add("auc", FilterAUC)
+
+
+auc = function (truth, prob) {
+  n_pos = sum(truth)
+  n_neg = length(truth) - n_pos
+  if (n_pos == 0L || n_neg == 0L)
+    return(0.5)
+  r = rank(prob, ties.method = "average")
+  (sum(r[truth]) - n_pos * (n_pos + 1L)/2L)/(n_pos * n_neg)
+}

README.md

@@ -58,7 +58,7 @@ as.data.table(filter$calculate(task))
 | Name              | Task Type      | Feature Types                                         | Package                                                           |
 | :---------------- | :------------- | :---------------------------------------------------- | :---------------------------------------------------------------- |
 | anova             | Classif        | Integer, Numeric                                      | stats                                                             |
-| auc               | Classif        | Integer, Numeric                                      | [Metrics](https://cran.r-project.org/package=Metrics)             |
+| auc               | Classif        | Integer, Numeric                                      | [mlr3measures](https://cran.r-project.org/package=mlr3measures)   |
 | carscore          | Regr           | Numeric                                               | [care](https://cran.r-project.org/package=care)                   |
 | cmim              | Classif & Regr | Integer, Numeric, Factor, Ordered                     | [praznik](https://cran.r-project.org/package=praznik)             |
 | correlation       | Regr           | Integer, Numeric                                      | stats                                                             |
@@ -79,8 +79,8 @@ as.data.table(filter$calculate(task))
 The following learners allow the extraction of variable importance and
 therefore are supported by `FilterImportance`:
 
-    ## [1] "classif.featureless" "classif.ranger"      "classif.rpart"      
-    ## [4] "classif.xgboost"     "regr.featureless"    "regr.ranger"        
+    ## [1] "classif.featureless" "classif.ranger"      "classif.rpart"
+    ## [4] "classif.xgboost"     "regr.featureless"    "regr.ranger"
     ## [7] "regr.rpart"          "regr.xgboost"
 
 If your learner is not listed here but capable of extracting variable