Changeset 10715 for branches/DataPreprocessing/HeuristicLab.DataPreprocessing

Timestamp:

04/02/14 15:48:53 (11 years ago)

Author:

mleitner

Message:

Add methods for preview of deleting samples/variables without sufficient information

Location:

branches/DataPreprocessing/HeuristicLab.DataPreprocessing/3.3

Files:

• Implementations/ManipulationLogic.cs (modified) (1 diff)
• Interfaces/IManipulationLogic.cs (modified) (1 diff)

branches/DataPreprocessing/HeuristicLab.DataPreprocessing/3.3/Implementations/ManipulationLogic.cs

@@ -257 +257 @@
 
 
+    public List<int> RowsWithMissingValuesGreater(double percent) {
+
+      List<int> rows= new List<int>();
+
+      for (int i = preprocessingData.Rows - 1; i >= 0; --i)
+      {
+        int missingCount = statisticsLogic.GetRowMissingValueCount(i);
+        if (100f / preprocessingData.Columns * missingCount >= percent)
+        {
+          rows.Add(i);
+        }
+      }
+
+      return rows;
+    }
+
+    public List<int> ColumnsWithMissingValuesGreater(double percent) {
+
+      List<int> columns = new List<int>();
+      for (int i = preprocessingData.Columns - 1; i >= 0; --i)
+      {
+        int missingCount = statisticsLogic.GetMissingValueCount(i);
+        if (100f / preprocessingData.Columns * missingCount >= percent)
+        {
+          columns.Add(i);
+        }
+      }
+
+      return columns;
+    }
+
+    public List<int> ColumnsWithVarianceSmaller(double variance) {
+
+      List<int> columns = new List<int>();
+      for (int i = preprocessingData.Columns - 1; i >= 0; --i)
+      {
+        if (preprocessingData.IsType<double>(i) || preprocessingData.IsType<DateTime>(i))
+        {
+          double columnVariance = statisticsLogic.GetVariance(i);
+          if (columnVariance < variance)
+          {
+            columns.Add(i);
+          }
+        }
+      }
+      return columns;
+    }
+
     public void DeleteRowsWithMissingValuesGreater(double percent) {
-      for (int i = 0; i < preprocessingData.Rows; ++i) {
-        int missingCount = statisticsLogic.GetRowMissingValueCount(i);
-        if (100f / preprocessingData.Columns * missingCount >= percent) {
-          preprocessingData.DeleteRow(i);
-          --i;
-        }
-      }
-    }
-
-    public void DeleteColumnsWithMissingValuesGreater(float percent) {
-      for (int i = 0; i < preprocessingData.Columns; ++i) {
-        int missingCount = statisticsLogic.GetMissingValueCount(i);
-        if (100f / preprocessingData.Columns * missingCount >= percent) {
-          preprocessingData.DeleteColumn(i);
-          --i;
-        }
-      }
+      DeleteRows(RowsWithMissingValuesGreater(percent));
+    }
+
+    public void DeleteColumnsWithMissingValuesGreater(double percent) {
+      DeleteColumns(ColumnsWithMissingValuesGreater(percent));
     }
 
     public void DeleteColumnsWithVarianceSmaller(double variance) {
-      for (int i = 0; i < preprocessingData.Columns; ++i) {
-        if (preprocessingData.IsType<double>(i) || preprocessingData.IsType<DateTime>(i)) {
-          double columnVariance = statisticsLogic.GetVariance(i);
-          if (columnVariance < variance) {
-            preprocessingData.DeleteColumn(i);
-            --i;
-          }
-        }
-      }
-    }
-
+      DeleteColumns(ColumnsWithVarianceSmaller(variance));
+    }
+
+    private void DeleteRows(IEnumerable<int> rows) {
+      preprocessingData.InTransaction(() =>
+      {
+        foreach (int row in rows)
+        {
+          preprocessingData.DeleteRow(row);
+        }
+      });
+    }
+
+    private void DeleteColumns(IEnumerable<int> columns) {
+      preprocessingData.InTransaction(() =>
+      {
+        foreach (int column in columns)
+        {
+          preprocessingData.DeleteColumn(column);
+        }
+      });
+    }
   }
 }

branches/DataPreprocessing/HeuristicLab.DataPreprocessing/3.3/Interfaces/IManipulationLogic.cs

@@ -36 +36 @@
     void ShuffleWithRanges();
     void ShuffleWithRanges(IEnumerable<HeuristicLab.Data.IntRange> ranges);
+    List<int> RowsWithMissingValuesGreater(double percent);
+    List<int> ColumnsWithMissingValuesGreater(double percent);
+    List<int> ColumnsWithVarianceSmaller(double variance);
     void DeleteRowsWithMissingValuesGreater(double percent);
-    void DeleteColumnsWithMissingValuesGreater(float percent);
+    void DeleteColumnsWithMissingValuesGreater(double percent);
     void DeleteColumnsWithVarianceSmaller(double variance);
   }