Changeset 10948 for branches/DataPreprocessing/HeuristicLab.DataPreprocessing

Timestamp:

06/04/14 14:57:53 (10 years ago)

Author:

pfleck

Message:

• Implemented preserve/copy columns on transformations.

File:

• branches/DataPreprocessing/HeuristicLab.DataPreprocessing/3.4/PreprocessingTransformator.cs (modified) (5 diffs)

branches/DataPreprocessing/HeuristicLab.DataPreprocessing/3.4/PreprocessingTransformator.cs

@@ -23 +23 @@
 using System.Collections.Generic;
 using System.Linq;
+using System.Text;
+using HeuristicLab.Data;
 using HeuristicLab.Problems.DataAnalysis.Transformations;
 
@@ -29 +31 @@
     private readonly ITransactionalPreprocessingData preprocessingData;
 
+    private readonly IDictionary<string, IList<double>> originalColumns;
+
     public PreprocessingTransformator(IPreprocessingData preprocessingData) {
       this.preprocessingData = (ITransactionalPreprocessingData)preprocessingData;
+      originalColumns = new Dictionary<string, IList<double>>();
     }
 
-    public bool ApplyTransformations(IEnumerable<ITransformation> transformations, out string errorMsg) {
+    public bool ApplyTransformations(IEnumerable<ITransformation> transformations, bool preserveColumns, out string errorMsg) {
       bool success;
 
@@ -39 +44 @@
       try {
         var doubleTransformations = transformations.OfType<Transformation<double>>().ToList();
-        ApplyDoubleTranformations(doubleTransformations, out success, out errorMsg);
+        ApplyDoubleTranformations(doubleTransformations, preserveColumns, out success, out errorMsg);
+
+        if (preserveColumns) {
+          RestorePreservedColumns(doubleTransformations);
+          originalColumns.Clear();
+        }
       } finally {
         preprocessingData.EndTransaction();
-        //if (!success)
-        //preprocessingData.Undo();
       }
 
@@ -49 +57 @@
     }
 
-    private void ApplyDoubleTranformations(IEnumerable<Transformation<double>> transformations, out bool success, out string errorMsg) {
+    private void ApplyDoubleTranformations(IList<Transformation<double>> transformations, bool preserveColumns, out bool success, out string errorMsg) {
       errorMsg = string.Empty;
       success = true;
@@ -56 +64 @@
 
         var originalData = preprocessingData.GetValues<double>(colIndex);
+
+        if (preserveColumns && !originalColumns.ContainsKey(transformation.Column))
+          originalColumns.Add(transformation.Column, originalData);
+
         string errorMsgPart;
         bool successPart;
         var transformedData = ApplyDoubleTransformation(transformation, originalData, out successPart, out errorMsgPart);
         errorMsg += errorMsgPart + Environment.NewLine;
-        //if (!success) return;
+
         if (!successPart) success = false;
         preprocessingData.SetValues(colIndex, transformedData.ToList());
         preprocessingData.Transformations.Add(transformation);
       }
-
     }
 
-    private IEnumerable<double> ApplyDoubleTransformation(Transformation<double> transformation, IEnumerable<double> data, out bool success, out string errorMsg) {
+    private IEnumerable<double> ApplyDoubleTransformation(Transformation<double> transformation, IList<double> data, out bool success, out string errorMsg) {
       success = transformation.Check(data, out errorMsg);
       return transformation.Apply(data);
     }
+
+    private void RestorePreservedColumns(IList<Transformation<double>> transformations) {
+      foreach (var column in originalColumns) {
+        int originalColumnIndex = preprocessingData.GetColumnIndex(column.Key);
+        int newColumnIndex = originalColumnIndex + 1;
+        string newColumnName = GetTransformatedColumnName(transformations, column.Key);
+
+        // create new transformed column
+        preprocessingData.InsertColumn<double>(newColumnName, newColumnIndex);
+        preprocessingData.SetValues(newColumnIndex, preprocessingData.GetValues<double>(originalColumnIndex));
+        // restore old values
+        preprocessingData.SetValues(originalColumnIndex, column.Value);
+      }
+    }
+
+    private string GetTransformatedColumnName(IList<Transformation<double>> transformations, string column) {
+      string suffix = GetTransformationSuffix(transformations, column);
+      return column + "_" + suffix;
+    }
+
+    private string GetTransformationSuffix(IList<Transformation<double>> transformations, string column) {
+      var suffixes = transformations.Where(t => t.Column == column).Select(t => t.ShortName);
+      var builder = new StringBuilder();
+      foreach (var suffix in suffixes) {
+        builder.Append(suffix);
+      }
+      return builder.ToString();
+    }
   }
 }