Changeset 10590 for branches/DataPreprocessing/HeuristicLab.DataPreprocessing/3.3/Implementations

Timestamp:

03/12/14 17:42:42 (11 years ago)

Author:

sbreuer

Message:

• enhanced selection of cells

File:

• branches/DataPreprocessing/HeuristicLab.DataPreprocessing/3.3/Implementations/ManipulationLogic.cs (modified) (2 diffs)

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

@@ -43 +43 @@
     }
 
-    public void ReplaceIndicesByAverageValue(int columnIndex, IEnumerable<int> rowIndices) {
-      double average = statisticInfo.GetAverage(columnIndex);
-      ReplaceIndicesByValue<double>(columnIndex, rowIndices, average);
-    }
-
-    public void ReplaceIndicesByMedianValue(int columnIndex, IEnumerable<int> rowIndices) {
-      double median = statisticInfo.GetMedian(columnIndex);
-      ReplaceIndicesByValue<double>(columnIndex, rowIndices, median);
-    }
-
-    public void ReplaceIndicesByRandomValue(int columnIndex, IEnumerable<int> rowIndices) {
+    public void ReplaceIndicesByAverageValue(Dictionary<int, List<int>> cells) {
+      preprocessingData.BeginTransaction();
+      foreach (var column in cells) {
+        double average = statisticInfo.GetAverage(column.Key);
+        ReplaceIndicesByValue<double>(column.Key, column.Value, average);
+      }
+      preprocessingData.EndTransaction();
+    }
+
+    public void ReplaceIndicesByMedianValue(Dictionary<int, List<int>> cells) {
+      preprocessingData.BeginTransaction();
+      foreach (var column in cells) {
+        double median = statisticInfo.GetMedian(column.Key);
+        ReplaceIndicesByValue<double>(column.Key, column.Value, median);
+      }
+      preprocessingData.EndTransaction();
+    }
+
+    public void ReplaceIndicesByRandomValue(Dictionary<int, List<int>> cells) {
+      preprocessingData.BeginTransaction();
       Random r = new Random();
 
-      double max = statisticInfo.GetMax<double>(columnIndex);
-      double min = statisticInfo.GetMin<double>(columnIndex);
-      double randMultiplier = (max - min);
-      foreach (int index in rowIndices) {
-        double rand = r.NextDouble() * randMultiplier + min;
-        preprocessingData.SetCell<double>(columnIndex, index, rand);
-      }
-    }
-
-    public void ReplaceIndicesByLinearInterpolationOfNeighbours(int columnIndex, IEnumerable<int> rowIndices) {
-      int countValues = preprocessingData.GetValues<double>(columnIndex).Count();
-      foreach (int index in rowIndices) {
-        // dont replace first or last values
-        if (index > 0 && index < countValues) {
-          int prevIndex = indexOfPrevPresentValue(columnIndex, index);
-          int nextIndex = indexOfNextPresentValue(columnIndex, index);
-
-          // no neighbours found
-          if (prevIndex < 0 && nextIndex >= countValues) {
-            continue;
+      foreach (var column in cells) {
+        double max = statisticInfo.GetMax<double>(column.Key);
+        double min = statisticInfo.GetMin<double>(column.Key);
+        double randMultiplier = (max - min);
+        foreach (int index in column.Value) {
+          double rand = r.NextDouble() * randMultiplier + min;
+          preprocessingData.SetCell<double>(column.Key, index, rand);
+        }
+      }
+      preprocessingData.EndTransaction();
+    }
+
+    public void ReplaceIndicesByLinearInterpolationOfNeighbours(Dictionary<int, List<int>> cells) {
+      preprocessingData.BeginTransaction();
+      foreach (var column in cells) {
+        int countValues = preprocessingData.GetValues<double>(column.Key).Count();
+        foreach (int index in column.Value) {
+          // dont replace first or last values
+          if (index > 0 && index < countValues) {
+            int prevIndex = indexOfPrevPresentValue(column.Key, index);
+            int nextIndex = indexOfNextPresentValue(column.Key, index);
+
+            // no neighbours found
+            if (prevIndex < 0 && nextIndex >= countValues) {
+              continue;
+            }
+            double prev = preprocessingData.GetCell<double>(column.Key, prevIndex);
+            double next = preprocessingData.GetCell<double>(column.Key, nextIndex);
+
+            int valuesToInterpolate = nextIndex - prevIndex;
+
+            double interpolationStep = (prev + next) / valuesToInterpolate;
+
+            for (int i = prevIndex; i < nextIndex; ++i) {
+              double interpolated = prev + (interpolationStep * (i - prevIndex));
+              preprocessingData.SetCell<double>(column.Key, i, interpolated);
+            }
           }
-          double prev = preprocessingData.GetCell<double>(columnIndex, prevIndex);
-          double next = preprocessingData.GetCell<double>(columnIndex, nextIndex);
-
-          int valuesToInterpolate = nextIndex - prevIndex;
-
-          double interpolationStep = (prev + next) / valuesToInterpolate;
-
-          for (int i = prevIndex; i < nextIndex; ++i) {
-            double interpolated = prev + (interpolationStep * (i - prevIndex));
-            preprocessingData.SetCell<double>(columnIndex, i, interpolated);
-          }
-        }
-      }
+        }
+      }
+      preprocessingData.EndTransaction();
     }
 
@@ -110 +126 @@
     }
 
-    public void ReplaceIndicesByMostCommonValue(int columnIndex, IEnumerable<int> rowIndices) {
-      if (preprocessingData.IsType<double>(columnIndex)) {
-        ReplaceIndicesByValue<double>(columnIndex, rowIndices, statisticInfo.GetMostCommonValue<double>(columnIndex));
-      } else if (preprocessingData.IsType<string>(columnIndex)) {
-        ReplaceIndicesByValue<string>(columnIndex, rowIndices, statisticInfo.GetMostCommonValue<string>(columnIndex));
-      } else if (preprocessingData.IsType<DateTime>(columnIndex)) {
-        ReplaceIndicesByValue<DateTime>(columnIndex, rowIndices, statisticInfo.GetMostCommonValue<DateTime>(columnIndex));
-      } else {
-        throw new ArgumentException("column with index: " + columnIndex + " contains a non supported type.");
-      }
+    public void ReplaceIndicesByMostCommonValue(Dictionary<int, List<int>> cells) {
+      preprocessingData.BeginTransaction();
+      foreach (var column in cells) {
+        if (preprocessingData.IsType<double>(column.Key)) {
+          ReplaceIndicesByValue<double>(column.Key, column.Value, statisticInfo.GetMostCommonValue<double>(column.Key));
+        } else if (preprocessingData.IsType<string>(column.Key)) {
+          ReplaceIndicesByValue<string>(column.Key, column.Value, statisticInfo.GetMostCommonValue<string>(column.Key));
+        } else if (preprocessingData.IsType<DateTime>(column.Key)) {
+          ReplaceIndicesByValue<DateTime>(column.Key, column.Value, statisticInfo.GetMostCommonValue<DateTime>(column.Key));
+        } else {
+          throw new ArgumentException("column with index: " + column.Key + " contains a non supported type.");
+        }
+      }
+      preprocessingData.EndTransaction();
     }