Changeset 14579 for branches/DataPreprocessing Enhancements/HeuristicLab.DataPreprocessing/3.4/Content

Timestamp:

01/18/17 11:17:19 (8 years ago)

Author:

mkommend

Message:

#2709: Refactored histogram view and content to support grouping by string and datetime variables.

File:

• branches/DataPreprocessing Enhancements/HeuristicLab.DataPreprocessing/3.4/Content/HistogramContent.cs (modified) (3 diffs)

branches/DataPreprocessing Enhancements/HeuristicLab.DataPreprocessing/3.4/Content/HistogramContent.cs

@@ -20 +20 @@
 #endregion
 
+using System;
 using System.Collections.Generic;
 using System.Drawing;
@@ -33 +34 @@
       get { return HeuristicLab.Common.Resources.VSImageLibrary.Statistics; }
     }
-    private const int MAX_DISTINCT_VALUES_FOR_CLASSIFCATION = 20;
 
-    public int ClassifierVariableIndex { get; set; }
+    private const int MAX_DISTINCT_VALUES_FOR_GROUPING = 20;
+
+    public string GroupingVariableName { get; set; }
 
     public int Bins { get; set; }
@@ -53 +55 @@
     }
 
-    public IEnumerable<string> GetVariableNamesForHistogramClassification() {
-      List<string> doubleVariableNames = new List<string>();
+    public IEnumerable<string> GetVariableNamesForGrouping() {
+      var variableNames = new List<string>();
 
-      //only return variable names from type double
       for (int i = 0; i < PreprocessingData.Columns; ++i) {
-        if (PreprocessingData.VariableHasType<double>(i)) {
-          double distinctValueCount = PreprocessingData.GetValues<double>(i).GroupBy(x => x).Count();
-          bool distinctValuesOk = distinctValueCount <= MAX_DISTINCT_VALUES_FOR_CLASSIFCATION;
-          if (distinctValuesOk)
-            doubleVariableNames.Add(PreprocessingData.GetVariableName(i));
-        }
+        int distinctValues = int.MaxValue;
+        if (PreprocessingData.VariableHasType<double>(i))
+          distinctValues = PreprocessingData.GetValues<double>(i).GroupBy(x => x).Count();
+        else if (PreprocessingData.VariableHasType<string>(i))
+          distinctValues = PreprocessingData.GetValues<string>(i).GroupBy(x => x).Count();
+        else if (PreprocessingData.VariableHasType<DateTime>(i))
+          distinctValues = PreprocessingData.GetValues<DateTime>(i).GroupBy(x => x).Count();
+
+        if (distinctValues <= MAX_DISTINCT_VALUES_FOR_GROUPING)
+          variableNames.Add(PreprocessingData.GetVariableName(i));
       }
-      return doubleVariableNames;
+      return variableNames;
     }
   }