Changeset 4202

Timestamp:

08/11/10 18:39:49 (14 years ago)

Author:

gkronber

Message:

Changed R² evaluator to return 0 when the estimated values contain NaN values. #1142.

Location:

branches/DataAnalysis/HeuristicLab.Problems.DataAnalysis.Regression/3.3/Symbolic

Files:

• Analyzers/FixedValidationBestScaledSymbolicRegressionSolutionAnalyzer.cs (modified) (3 diffs)
• Evaluators/SymbolicRegressionPearsonsRSquaredEvaluator.cs (modified) (1 diff)

branches/DataAnalysis/HeuristicLab.Problems.DataAnalysis.Regression/3.3/Symbolic/Analyzers/FixedValidationBestScaledSymbolicRegressionSolutionAnalyzer.cs

@@ -32 +32 @@
 using HeuristicLab.Problems.DataAnalysis.Evaluators;
 using HeuristicLab.Problems.DataAnalysis.Symbolic;
+using System;
 
 namespace HeuristicLab.Problems.DataAnalysis.Regression.Symbolic.Analyzers {
@@ -285 +286 @@
         int trainingEnd = ProblemData.TrainingSamplesEnd.Value;
         IEnumerable<int> trainingRows = Enumerable.Range(trainingStart, trainingEnd - trainingStart);
-        SymbolicRegressionScaledMeanSquaredErrorEvaluator.Calculate(SymbolicExpressionTreeInterpreter, bestTree,
-          lowerEstimationLimit, upperEstimationLimit,
-          ProblemData.Dataset, targetVariable,
-          trainingRows, out beta, out alpha);
+        IEnumerable<double> originalValues = ProblemData.Dataset.GetEnumeratedVariableValues(targetVariable, trainingRows);
+        IEnumerable<double> estimatedValues =
+          GetBoundedValues(SymbolicExpressionTreeInterpreter.GetSymbolicExpressionTreeValues(bestTree, ProblemData.Dataset, trainingRows),
+            lowerEstimationLimit, upperEstimationLimit);
+
+
+        SymbolicRegressionScaledMeanSquaredErrorEvaluator.CalculateScalingParameters(originalValues, estimatedValues, out beta, out alpha);
 
         // scale tree for solution
@@ -317 +321 @@
       AddValue(validationValues, bestQuality, CurrentBestValidationQualityParameterName, CurrentBestValidationQualityParameterName);
       return base.Apply();
+    }
+
+    private IEnumerable<double> GetBoundedValues(IEnumerable<double> values, double lowerEstimationLimit, double upperEstimationLimit) {
+      foreach (double v in values) {
+        if (double.IsNaN(v)) yield return upperEstimationLimit;
+        else yield return Math.Min(upperEstimationLimit, Math.Max(lowerEstimationLimit, v));
+      }
     }
 

branches/DataAnalysis/HeuristicLab.Problems.DataAnalysis.Regression/3.3/Symbolic/Evaluators/SymbolicRegressionPearsonsRSquaredEvaluator.cs

@@ -54 +54 @@
         double original = originalEnumerator.Current;
         if (double.IsNaN(estimated))
-          estimated = upperEstimationLimit;
+          return 0.0;
         else
           estimated = Math.Min(upperEstimationLimit, Math.Max(lowerEstimationLimit, estimated));