Changeset 10218 for branches/DataPreprocessing/HeuristicLab.DataPreprocessing

Timestamp:

12/11/13 15:55:30 (10 years ago)

Author:

mleitner

Message:

Implement first draft for shuffeling dataset while maintaining Test- and Trainingspartition

Location:

branches/DataPreprocessing/HeuristicLab.DataPreprocessing/3.3

Files:

• Implementations/PreprocessingData.cs (modified) (3 diffs)
• Implementations/PreprocessingDataManipulation.cs (modified) (2 diffs)
• Interfaces/IPreprocessingData.cs (modified) (2 diffs)

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

@@ -27 +27 @@
 using HeuristicLab.Core;
 using HeuristicLab.Problems.DataAnalysis;
+using HeuristicLab.Data;
 
 namespace HeuristicLab.DataPreprocessing {
@@ -35 +36 @@
 
     private IList<string> variableNames;
+    private IntRange trainingPartition;
+    private IntRange testPartition;
 
     private IDictionary<string, int> variableNameIndices;
@@ -71 +74 @@
         }
       }
+
+      trainingPartition = problemData.TrainingPartition;
+      testPartition = problemData.TestPartition;
 
       trainingToTestRatio = (double)problemData.TrainingPartition.Size / problemData.TestPartition.Size;

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

@@ -1 @@
-using System;
+using HeuristicLab.Data;
+using System;
 using System.Collections.Generic;
 using System.Linq;
@@ -87 +88 @@
             }
         }
+
+        public void ShuffleWithRanges(IEnumerable<IntRange> ranges)
+        {
+            // init random outside loop
+            Random random = new Random();
+
+            // process all given ranges - e.g. TrainingPartition, Trainingpartition
+            foreach (IntRange range in ranges) { 
+                List<int> shuffledIndices = new List<int>();
+               
+                // generate random indices used for shuffeling each column
+                for (int i = range.End; i > range.Start; --i)
+                {
+                    int rand = random.Next(range.Start, i);
+                    shuffledIndices[i] = rand;
+                }
+
+                foreach (string variableName in preprocessingData.VariableNames)
+                {
+                    if (preprocessingData.IsType<double>(variableName))
+                    {
+                        reOrderToIndices<double>(variableName, shuffledIndices);
+                    }
+                    else if (preprocessingData.IsType<string>(variableName))
+                    {
+                        reOrderToIndices<string>(variableName, shuffledIndices);
+                    }
+                    else if (preprocessingData.IsType<DateTime>(variableName))
+                    {
+                        reOrderToIndices<DateTime>(variableName, shuffledIndices);
+                    }
+                }
+            }      
+        }
+
+        public void reOrderToIndices<T>(string variableName, List<int> indices) {
+            // process all columns equally
+            for (int i = 0; i < preprocessingData.Rows; i++)
+            {
+                int replaceIndex = indices[i];
+
+                T tmp = preprocessingData.GetCell<T>(variableName, i);
+                T replaceValue = preprocessingData.GetCell<T>(variableName, replaceIndex);
+
+                preprocessingData.SetCell<T>(variableName, i, replaceValue);
+                preprocessingData.SetCell<T>(variableName, replaceIndex, tmp);
+            }
+        }
     }
 }

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

@@ -23 +23 @@
 using HeuristicLab.Core;
 using HeuristicLab.Problems.DataAnalysis;
+using HeuristicLab.Data;
 
 namespace HeuristicLab.DataPreprocessing {
@@ -35 +36 @@
     void InsertRow(int rowIndex);
     void DeleteRow(int rowIndex);
+
+    IntRange TrainingPartition { get; }
+    IntRange TestPartition { get; }
 
     void InsertColumn<T>(string variableName, int columnIndex);