Changeset 3806 for trunk/sources/HeuristicLab.Problems.DataAnalysis.Regression

Timestamp:

05/14/10 14:17:07 (15 years ago)

Author:

gkronber

Message:

Fixed a bug in symbolic regression analyzers. (MSE, R² and relative errors were evaluated on the non-scaled solution). #938

Location:

trunk/sources/HeuristicLab.Problems.DataAnalysis.Regression/3.3/Symbolic

Files:

• Analyzers/ValidationBestScaledSymbolicRegressionSolutionAnalyzer.cs (modified) (9 diffs)
• SymbolicRegressionProblem.cs (modified) (3 diffs)

trunk/sources/HeuristicLab.Problems.DataAnalysis.Regression/3.3/Symbolic/Analyzers/ValidationBestScaledSymbolicRegressionSolutionAnalyzer.cs

@@ -49 +49 @@
     private const string SymbolicExpressionTreeInterpreterParameterName = "SymbolicExpressionTreeInterpreter";
     private const string ProblemDataParameterName = "ProblemData";
-    private const string SamplesStartParameterName = "SamplesStart";
-    private const string SamplesEndParameterName = "SamplesEnd";
+    private const string TrainingSamplesStartParameterName = "TrainingSamplesStart";
+    private const string TrainingSamplesEndParameterName = "TrainingSamplesEnd";
+    private const string ValidationSamplesStartParameterName = "ValidationSamplesStart";
+    private const string ValidationSamplesEndParameterName = "ValidationSamplesEnd";
+    private const string TestSamplesStartParameterName = "TestSamplesStart";
+    private const string TestSamplesEndParameterName = "TestSamplesEnd";
     private const string QualityParameterName = "Quality";
     private const string ScaledQualityParameterName = "ScaledQuality";
@@ -75 +79 @@
       get { return (IValueLookupParameter<DataAnalysisProblemData>)Parameters[ProblemDataParameterName]; }
     }
-    public IValueLookupParameter<IntValue> SamplesStartParameter {
-      get { return (IValueLookupParameter<IntValue>)Parameters[SamplesStartParameterName]; }
-    }
-    public IValueLookupParameter<IntValue> SamplesEndParameter {
-      get { return (IValueLookupParameter<IntValue>)Parameters[SamplesEndParameterName]; }
+    public IValueLookupParameter<IntValue> TrainingSamplesStartParameter {
+      get { return (IValueLookupParameter<IntValue>)Parameters[TrainingSamplesStartParameterName]; }
+    }
+    public IValueLookupParameter<IntValue> TrainingSamplesEndParameter {
+      get { return (IValueLookupParameter<IntValue>)Parameters[TrainingSamplesEndParameterName]; }
+    }
+    public IValueLookupParameter<IntValue> ValidationSamplesStartParameter {
+      get { return (IValueLookupParameter<IntValue>)Parameters[ValidationSamplesStartParameterName]; }
+    }
+    public IValueLookupParameter<IntValue> ValidationSamplesEndParameter {
+      get { return (IValueLookupParameter<IntValue>)Parameters[ValidationSamplesEndParameterName]; }
+    }
+    public IValueLookupParameter<IntValue> TestSamplesStartParameter {
+      get { return (IValueLookupParameter<IntValue>)Parameters[TestSamplesStartParameterName]; }
+    }
+    public IValueLookupParameter<IntValue> TestSamplesEndParameter {
+      get { return (IValueLookupParameter<IntValue>)Parameters[TestSamplesEndParameterName]; }
     }
     public IValueLookupParameter<DoubleValue> UpperEstimationLimitParameter {
@@ -108 +124 @@
     [Storable]
     private BestQualityMemorizer bestKnownQualityMemorizer;
+    [Storable]
+    private SymbolicRegressionModelQualityAnalyzer modelQualityAnalyzer;
 
     public ValidationBestScaledSymbolicRegressionSolutionAnalyzer()
@@ -115 +133 @@
       Parameters.Add(new ValueLookupParameter<ISymbolicExpressionTreeInterpreter>(SymbolicExpressionTreeInterpreterParameterName, "The interpreter that should be used for the analysis of symbolic expression trees."));
       Parameters.Add(new ValueLookupParameter<DataAnalysisProblemData>(ProblemDataParameterName, "The problem data for which the symbolic expression tree is a solution."));
-      Parameters.Add(new ValueLookupParameter<IntValue>(SamplesStartParameterName, "The first index of the validation partition of the data set."));
-      Parameters.Add(new ValueLookupParameter<IntValue>(SamplesEndParameterName, "The last index of the validation partition of the data set."));
+      Parameters.Add(new ValueLookupParameter<IntValue>(TrainingSamplesStartParameterName, "The first index of the training partition of the data set."));
+      Parameters.Add(new ValueLookupParameter<IntValue>(TrainingSamplesEndParameterName, "The last index of the training partition of the data set."));
+      Parameters.Add(new ValueLookupParameter<IntValue>(ValidationSamplesStartParameterName, "The first index of the validation partition of the data set."));
+      Parameters.Add(new ValueLookupParameter<IntValue>(ValidationSamplesEndParameterName, "The last index of the validation partition of the data set."));
+      Parameters.Add(new ValueLookupParameter<IntValue>(TestSamplesStartParameterName, "The first index of the test partition of the data set."));
+      Parameters.Add(new ValueLookupParameter<IntValue>(TestSamplesEndParameterName, "The last index of the test partition of the data set."));
       Parameters.Add(new ValueLookupParameter<DoubleValue>(UpperEstimationLimitParameterName, "The upper estimation limit that was set for the evaluation of the symbolic expression trees."));
       Parameters.Add(new ValueLookupParameter<DoubleValue>(LowerEstimationLimitParameterName, "The lower estimation limit that was set for the evaluation of the symbolic expression trees."));
@@ -127 +149 @@
       subScopesProcessor = new UniformSubScopesProcessor();
       SymbolicRegressionSolutionLinearScaler linearScaler = new SymbolicRegressionSolutionLinearScaler();
+      modelQualityAnalyzer = new SymbolicRegressionModelQualityAnalyzer();
       SymbolicRegressionMeanSquaredErrorEvaluator validationMseEvaluator = new SymbolicRegressionMeanSquaredErrorEvaluator();
       bestSolutionAnalyzer = new BestSymbolicRegressionSolutionAnalyzer();
@@ -143 +166 @@
       linearScaler.ScaledSymbolicExpressionTreeParameter.ActualName = ScaledSymbolicExpressionTreeParameterName;
 
+      modelQualityAnalyzer.ProblemDataParameter.ActualName = ProblemDataParameter.Name;
+      modelQualityAnalyzer.SymbolicExpressionTreeParameter.ActualName = ScaledSymbolicExpressionTreeParameterName;
+      modelQualityAnalyzer.SymbolicExpressionTreeParameter.Depth = SymbolicExpressionTreeParameter.Depth;
+      modelQualityAnalyzer.UpperEstimationLimitParameter.ActualName = UpperEstimationLimitParameter.Name;
+      modelQualityAnalyzer.LowerEstimationLimitParameter.ActualName = LowerEstimationLimitParameter.Name;
+      modelQualityAnalyzer.SymbolicExpressionTreeInterpreterParameter.ActualName = SymbolicExpressionTreeInterpreterParameter.Name;
+      modelQualityAnalyzer.TrainingSamplesStartParameter.ActualName = TrainingSamplesStartParameter.Name;
+      modelQualityAnalyzer.TrainingSamplesEndParameter.ActualName = TrainingSamplesEndParameter.Name;
+      modelQualityAnalyzer.TestSamplesStartParameter.ActualName = TestSamplesStartParameter.Name;
+      modelQualityAnalyzer.TestSamplesEndParameter.ActualName = TestSamplesEndParameter.Name;
+
       validationMseEvaluator.LowerEstimationLimitParameter.ActualName = LowerEstimationLimitParameter.Name;
       validationMseEvaluator.UpperEstimationLimitParameter.ActualName = UpperEstimationLimitParameter.Name;
@@ -149 +183 @@
       validationMseEvaluator.QualityParameter.ActualName = ScaledQualityParameterName;
       validationMseEvaluator.RegressionProblemDataParameter.ActualName = ProblemDataParameter.Name;
-      validationMseEvaluator.SamplesStartParameter.ActualName = SamplesStartParameter.Name;
-      validationMseEvaluator.SamplesEndParameter.ActualName = SamplesEndParameter.Name;
+      validationMseEvaluator.SamplesStartParameter.ActualName = ValidationSamplesStartParameter.Name;
+      validationMseEvaluator.SamplesEndParameter.ActualName = ValidationSamplesEndParameter.Name;
 
       bestSolutionAnalyzer.BestSolutionParameter.ActualName = BestSolutionParameter.Name;
@@ -190 +224 @@
       linearScaler.Successor = validationMseEvaluator;
       validationMseEvaluator.Successor = null;
-      subScopesProcessor.Successor = bestSolutionAnalyzer;
+      subScopesProcessor.Successor = modelQualityAnalyzer;
+      modelQualityAnalyzer.Successor = bestSolutionAnalyzer;
       bestSolutionAnalyzer.Successor = bestAvgWorstValidationQualityCalculator;
       bestAvgWorstValidationQualityCalculator.Successor = bestKnownQualityMemorizer;
@@ -221 +256 @@
       bestAvgWorstValidationQualityCalculator.QualityParameter.Depth = SymbolicExpressionTreeParameter.Depth;
       bestKnownQualityMemorizer.QualityParameter.Depth = SymbolicExpressionTreeParameter.Depth;
+      modelQualityAnalyzer.SymbolicExpressionTreeParameter.Depth = SymbolicExpressionTreeParameter.Depth;
     }
   }

trunk/sources/HeuristicLab.Problems.DataAnalysis.Regression/3.3/Symbolic/SymbolicRegressionProblem.cs

@@ -371 +371 @@
       operators.Add(new MinAvgMaxSymbolicExpressionTreeSizeAnalyzer());
       operators.Add(new SymbolicRegressionVariableFrequencyAnalyzer());
-      operators.Add(new SymbolicRegressionModelQualityAnalyzer());
       ParameterizeOperators();
       ParameterizeAnalyzers();
@@ -401 +400 @@
           bestValidationSolutionAnalyzer.SymbolicExpressionTreeInterpreterParameter.ActualName = SymbolicExpressionTreeInterpreterParameter.Name;
           bestValidationSolutionAnalyzer.SymbolicExpressionTreeParameter.ActualName = SolutionCreator.SymbolicExpressionTreeParameter.ActualName;
-          bestValidationSolutionAnalyzer.SamplesStartParameter.Value = ValidationSamplesStart;
-          bestValidationSolutionAnalyzer.SamplesEndParameter.Value = ValidationSamplesEnd;
+          bestValidationSolutionAnalyzer.TrainingSamplesStartParameter.Value = TrainingSamplesStart;
+          bestValidationSolutionAnalyzer.TrainingSamplesEndParameter.Value = TrainingSamplesEnd;
+          bestValidationSolutionAnalyzer.ValidationSamplesStartParameter.Value = ValidationSamplesStart;
+          bestValidationSolutionAnalyzer.ValidationSamplesEndParameter.Value = ValidationSamplesEnd;
+          bestValidationSolutionAnalyzer.TestSamplesStartParameter.Value = TestSamplesStart;
+          bestValidationSolutionAnalyzer.TestSamplesEndParameter.Value = TestSamplesEnd;
           bestValidationSolutionAnalyzer.BestKnownQualityParameter.ActualName = BestKnownQualityParameter.Name;
           bestValidationSolutionAnalyzer.QualityParameter.ActualName = Evaluator.QualityParameter.ActualName;
@@ -410 +413 @@
           varFreqAnalyzer.ProblemDataParameter.ActualName = DataAnalysisProblemDataParameter.Name;
         }
-        var modelQualityAnalyzer = analyzer as SymbolicRegressionModelQualityAnalyzer;
-        if (modelQualityAnalyzer != null) {
-          modelQualityAnalyzer.ProblemDataParameter.ActualName = DataAnalysisProblemDataParameter.Name;
-          modelQualityAnalyzer.SymbolicExpressionTreeParameter.ActualName = SolutionCreator.SymbolicExpressionTreeParameter.ActualName;
-          modelQualityAnalyzer.UpperEstimationLimitParameter.ActualName = UpperEstimationLimitParameter.Name;
-          modelQualityAnalyzer.LowerEstimationLimitParameter.ActualName = LowerEstimationLimitParameter.Name;
-          modelQualityAnalyzer.SymbolicExpressionTreeInterpreterParameter.ActualName = SymbolicExpressionTreeInterpreterParameter.Name;
-          modelQualityAnalyzer.TrainingSamplesStartParameter.Value = TrainingSamplesStart;
-          modelQualityAnalyzer.TrainingSamplesEndParameter.Value = ValidationSamplesEnd;
-          modelQualityAnalyzer.TestSamplesStartParameter.Value = TestSamplesStart;
-          modelQualityAnalyzer.TestSamplesEndParameter.Value = TestSamplesEnd;
-        }
       }
       foreach (ISymbolicExpressionTreeAnalyzer analyzer in Operators.OfType<ISymbolicExpressionTreeAnalyzer>()) {