Changeset 10193

Timestamp:

12/04/13 16:52:31 (11 years ago)

Author:

mleitner

Message:

Move attribute replacement methods to separate class.

Location:

branches/DataPreprocessing/HeuristicLab.DataPreprocessing/3.3

Files:

• HeuristicLab.DataPreprocessing-3.3.csproj (modified) (1 diff)
• Implementations/PreprocessingData.cs (modified) (3 diffs)
• Implementations/PreprocessingDataManipulation.cs (added)

branches/DataPreprocessing/HeuristicLab.DataPreprocessing/3.3/HeuristicLab.DataPreprocessing-3.3.csproj

@@ -85 +85 @@
     </Compile>
     <Compile Include="Implementations\PreprocessingData.cs" />
+    <Compile Include="Implementations\PreprocessingDataManipulation.cs" />
     <Compile Include="Implementations\StatisticInfo.cs" />
     <Compile Include="Interfaces\IStatisticInfo.cs" />

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

@@ -39 +39 @@
 
     private double trainingToTestRatio;
-    private StatisticInfo statisticInfo;
-
+  
     private PreprocessingData(PreprocessingData original, Cloner cloner)
       : base(original, cloner) {
@@ -76 +75 @@
       Columns = problemData.Dataset.Columns;
       Rows = problemData.Dataset.Rows;
-
-      statisticInfo = new StatisticInfo(this);
     }
 
@@ -184 +181 @@
     }
 
-    public void ReplaceIndicesByValue<T>(string variableName, IEnumerable<int> indices, T value)
-    {
-        foreach (int index in indices)
-        {
-            SetCell<T>(variableName, index, value);
-        }
-    }
-
-    public void ReplaceIndicesByAverageValue(string variableName, IEnumerable<int> indices) {
-        double average = statisticInfo.GetAverage(variableName);
-        ReplaceIndicesByValue<double>(variableName, indices, average);
-    }
-
-    public void ReplaceIndicesByAverageValue(string variableName, IEnumerable<int> indices) {
-        double median = statisticInfo.GetMedian(variableName);
-        ReplaceIndicesByValue<double>(variableName, indices, median);
-    }
-
-    public void ReplaceIndicesByRandomValue(string variableName, IEnumerable<int> indices) {
-        Random r = new Random();
-
-        double max = statisticInfo.GetMax<double>(variableName);
-        double min = statisticInfo.GetMin<double>(variableName);
-        double randMultiplier = (max - min);
-        foreach (int index in indices)
-        {
-            double rand = r.NextDouble() * randMultiplier + min;
-            SetCell<double>(variableName, index, rand);
-        } 
-    }
-
-    public void ReplaceIndicesByLinearInterpolationOfNeighbours(string variableName, IEnumerable<int> indices) {
-        int countValues = GetValues<double>(variableName).Count();
-        foreach (int index in indices)
-        {
-            // dont replace first or last values
-            if (index > 0 && index < countValues) {
-               double prev = GetCell<double>(variableName, index - 1);
-               double next = GetCell<double>(variableName, index + 1);
-
-               double interpolated = (prev + next) / 2;
-
-               SetCell<double>(variableName, index, interpolated);
-            }
-        } 
-    }
-
-    public void ReplaceIndicesByMostCommonValue(string variableName, IEnumerable<int> indices) {   
-      if (IsType<double>(variableName)) {
-        ReplaceIndicesByValue<double>(variableName, indices,statisticInfo.GetMostCommonValue<double>(variableName));
-      } else if (IsType<string>(variableName)) {
-        ReplaceIndicesByValue<string>(variableName, indices, statisticInfo.GetMostCommonValue<string>(variableName));
-      } else if (IsType<DateTime>(variableName)) {
-        ReplaceIndicesByValue<DateTime>(variableName, indices, statisticInfo.GetMostCommonValue<DateTime>(variableName));
-      } else {
-        throw new ArgumentException("column with index: " + variableName + " contains a non supported type.");
-      }
-    }
-
     #endregion
   }