Changeset 9451

Timestamp:

05/06/13 17:18:05 (11 years ago)

Author:

abeham

Message:

#1913: fixed handling of non-sequential natural numbered class values in LDA initializer

File:

• trunk/sources/HeuristicLab.Algorithms.DataAnalysis/3.4/Nca/Initialization/LdaInitializer.cs (modified) (1 diff)

trunk/sources/HeuristicLab.Algorithms.DataAnalysis/3.4/Nca/Initialization/LdaInitializer.cs

@@ -48 +48 @@
                                                 data.TrainingIndices);
 
-      var uniqueClasses = data.Dataset.GetDoubleValues(data.TargetVariable, data.TrainingIndices).Distinct().Count();
+      // map class values to sequential natural numbers (required by alglib)
+      var uniqueClasses = data.Dataset.GetDoubleValues(data.TargetVariable, data.TrainingIndices)
+                                        .Distinct()
+                                        .Select((v, i) => new { v, i })
+                                        .ToDictionary(x => x.v, x => x.i);
+
+      for (int row = 0; row < instances; row++)
+        ldaDs[row, attributes] = uniqueClasses[ldaDs[row, attributes]];
 
       int info;
       double[,] matrix;
-      alglib.fisherldan(ldaDs, instances, attributes, uniqueClasses, out info, out matrix);
+      alglib.fisherldan(ldaDs, instances, attributes, uniqueClasses.Count, out info, out matrix);
 
       return matrix;