Changeset 8814 for branches/ClassificationEnsembleVoting/HeuristicLab.Problems.DataAnalysis/3.4/Implementation/Classification/WeightCalculators/MajorityVoteWeightCalculator.cs

Timestamp:

10/16/12 15:19:40 (12 years ago)

Author:

sforsten

Message:

#1776:

• improved performance of confidence calculation
• fixed bug in median confidence calculation
• fixed bug in average confidence calculation
• confidence calculation is now easier for training and test
• removed obsolete view ClassificationEnsembleSolutionConfidenceAccuracyDependence

File:

• branches/ClassificationEnsembleVoting/HeuristicLab.Problems.DataAnalysis/3.4/Implementation/Classification/WeightCalculators/MajorityVoteWeightCalculator.cs (modified) (1 diff)

branches/ClassificationEnsembleVoting/HeuristicLab.Problems.DataAnalysis/3.4/Implementation/Classification/WeightCalculators/MajorityVoteWeightCalculator.cs

@@ -52 +52 @@
     }
 
-    public override double GetConfidence(IEnumerable<IClassificationSolution> solutions, int index, double estimatedClassValue) {
+    public override double GetConfidence(IEnumerable<IClassificationSolution> solutions, int index, double estimatedClassValue, CheckPoint handler) {
       if (solutions.Count() < 1)
         return double.NaN;
-      Dataset dataset = solutions.First().ProblemData.Dataset;
-      int correctEstimated = solutions.Select(s => s.Model.GetEstimatedClassValues(dataset, Enumerable.Repeat(index, 1)).First())
+      var votingSolutions = solutions.Where(s => handler(s.ProblemData, index));
+      if (votingSolutions.Count() < 1)
+        return double.NaN;
+      Dataset dataset = votingSolutions.First().ProblemData.Dataset;
+      int correctEstimated = votingSolutions.Select(s => s.Model.GetEstimatedClassValues(dataset, Enumerable.Repeat(index, 1)).First())
                                       .Where(x => x.Equals(estimatedClassValue))
                                       .Count();
-      return ((double)correctEstimated / (double)solutions.Count() - 0.5) * 2;
+      return ((double)correctEstimated / (double)votingSolutions.Count() - 0.5) * 2;
     }
 
-    public override IEnumerable<double> GetConfidence(IEnumerable<IClassificationSolution> solutions, IEnumerable<int> indices, IEnumerable<double> estimatedClassValue) {
+    public override IEnumerable<double> GetConfidence(IEnumerable<IClassificationSolution> solutions, IEnumerable<int> indices, IEnumerable<double> estimatedClassValue, CheckPoint handler) {
       if (solutions.Count() < 1)
         return Enumerable.Repeat(double.NaN, indices.Count());
+
+      List<int> indicesList = indices.ToList();
       Dataset dataset = solutions.First().ProblemData.Dataset;
-      var estimationsPerSolution = solutions.Select(s => s.Model.GetEstimatedClassValues(dataset, indices).ToArray()).ToArray();
+      var solValues = solutions.ToDictionary(x => x, x => x.Model.GetEstimatedClassValues(dataset, indicesList).ToArray());
       double[] estimatedClassValueArr = estimatedClassValue.ToArray();
       double correctEstimated;
       double[] confidences = new double[indices.Count()];
 
-      for (int i = 0; i < indices.Count(); i++) {
-        correctEstimated = estimationsPerSolution.Where(x => DoubleExtensions.IsAlmost(x[i], estimatedClassValueArr[i])).Count();
-        confidences[i] = (correctEstimated / (double)solutions.Count() - 0.5) * 2;
+      for (int i = 0; i < indicesList.Count; i++) {
+        var votingSolutions = solValues.Where(x => handler(x.Key.ProblemData, indicesList[i]));
+        correctEstimated = votingSolutions.Where(x => DoubleExtensions.IsAlmost(x.Value[i], estimatedClassValueArr[i])).Count();
+        confidences[i] = (correctEstimated / (double)votingSolutions.Count() - 0.5) * 2;
       }