Changeset 13935 for trunk/sources/HeuristicLab.DataPreprocessing/3.4/Logic

Timestamp:

06/24/16 14:00:28 (8 years ago)

Author:

mkommend

Message:

#2616: Added default value for min, max and most common value operations for data preprocessing.

Location:

trunk/sources/HeuristicLab.DataPreprocessing/3.4/Logic

Files:

• ManipulationLogic.cs (modified) (3 diffs)
• StatisticsLogic.cs (modified) (3 diffs)

trunk/sources/HeuristicLab.DataPreprocessing/3.4/Logic/ManipulationLogic.cs

@@ -85 +85 @@
         foreach (var column in cells) {
           if (preprocessingData.VariableHasType<double>(column.Key)) {
-            double max = statisticsLogic.GetMax<double>(column.Key, considerSelection);
-            double min = statisticsLogic.GetMin<double>(column.Key, considerSelection);
+            double max = statisticsLogic.GetMax<double>(column.Key, double.NaN, considerSelection);
+            double min = statisticsLogic.GetMin<double>(column.Key, double.NaN, considerSelection);
             double randMultiplier = (max - min);
             foreach (int index in column.Value) {
@@ -93 +93 @@
             }
           } else if (preprocessingData.VariableHasType<DateTime>(column.Key)) {
-            DateTime min = statisticsLogic.GetMin<DateTime>(column.Key, considerSelection);
-            DateTime max = statisticsLogic.GetMax<DateTime>(column.Key, considerSelection);
+            DateTime min = statisticsLogic.GetMin<DateTime>(column.Key, DateTime.MinValue, considerSelection);
+            DateTime max = statisticsLogic.GetMax<DateTime>(column.Key, DateTime.MinValue, considerSelection);
             double randMultiplier = (max - min).TotalSeconds;
             foreach (int index in column.Value) {
@@ -213 +213 @@
         foreach (var column in cells) {
           if (preprocessingData.VariableHasType<double>(column.Key)) {
-            ReplaceIndicesByValue<double>(column.Key, column.Value, statisticsLogic.GetMostCommonValue<double>(column.Key, considerSelection));
+            ReplaceIndicesByValue<double>(column.Key, column.Value, statisticsLogic.GetMostCommonValue<double>(column.Key, double.NaN, considerSelection));
           } else if (preprocessingData.VariableHasType<string>(column.Key)) {
-            ReplaceIndicesByValue<string>(column.Key, column.Value, statisticsLogic.GetMostCommonValue<string>(column.Key, considerSelection));
+            ReplaceIndicesByValue<string>(column.Key, column.Value, statisticsLogic.GetMostCommonValue<string>(column.Key, string.Empty, considerSelection));
           } else if (preprocessingData.VariableHasType<DateTime>(column.Key)) {
-            ReplaceIndicesByValue<DateTime>(column.Key, column.Value, statisticsLogic.GetMostCommonValue<DateTime>(column.Key, considerSelection));
+            ReplaceIndicesByValue<DateTime>(column.Key, column.Value, statisticsLogic.GetMostCommonValue<DateTime>(column.Key, DateTime.MinValue, considerSelection));
           } else {
             throw new ArgumentException("column with index: " + column.Key + " contains a non supported type.");

trunk/sources/HeuristicLab.DataPreprocessing/3.4/Logic/StatisticsLogic.cs

@@ -71 +71 @@
     }
 
-    public T GetMin<T>(int columnIndex, bool considerSelection = false) where T : IComparable<T> {
-      var min = default(T);
+    public T GetMin<T>(int columnIndex, T defaultValue, bool considerSelection = false) where T : IComparable<T> {
+      var min = defaultValue;
       if (preprocessingData.VariableHasType<T>(columnIndex)) {
         var values = GetValuesWithoutNaN<T>(columnIndex, considerSelection);
@@ -82 +82 @@
     }
 
-    public T GetMax<T>(int columnIndex, bool considerSelection = false) where T : IComparable<T> {
-      var max = default(T);
+    public T GetMax<T>(int columnIndex, T defaultValue, bool considerSelection = false) where T : IComparable<T> {
+      var max = defaultValue;
       if (preprocessingData.VariableHasType<T>(columnIndex)) {
         var values = GetValuesWithoutNaN<T>(columnIndex, considerSelection);
@@ -131 +131 @@
     }
 
-    public T GetMostCommonValue<T>(int columnIndex, bool considerSelection = false) {
+    public T GetMostCommonValue<T>(int columnIndex, T defaultValue, bool considerSelection = false) {
       var values = GetValuesWithoutNaN<T>(columnIndex, considerSelection);
       if (!values.Any())
-        return default(T);
+        return defaultValue;
       return values.GroupBy(x => x)
                               .OrderByDescending(g => g.Count())