Changeset 5882 for trunk/sources/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Analyzers

Timestamp:

03/30/11 13:14:12 (14 years ago)

Author:

gkronber

Message:

#1418 renamed parameter and updated all validation analyzers to leave out test samples if the validation partition overlaps with the test partition.

Location:

trunk/sources/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Analyzers

Files:

• SymbolicDataAnalysisAnalyzer.cs (modified) (1 diff)
• SymbolicDataAnalysisMultiObjectiveTrainingBestSolutionAnalyzer.cs (modified) (1 diff)
• SymbolicDataAnalysisMultiObjectiveValidationAnalyzer.cs (modified) (5 diffs)
• SymbolicDataAnalysisMultiObjectiveValidationBestSolutionAnalyzer.cs (modified) (2 diffs)
• SymbolicDataAnalysisSingleObjectiveTrainingBestSolutionAnalyzer.cs (modified) (1 diff)
• SymbolicDataAnalysisSingleObjectiveValidationAnalyzer.cs (modified) (6 diffs)
• SymbolicDataAnalysisSingleObjectiveValidationBestSolutionAnalyzer.cs (modified) (1 diff)

trunk/sources/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Analyzers/SymbolicDataAnalysisAnalyzer.cs

@@ -49 +49 @@
     #endregion
     #region properties
-    public ItemArray<ISymbolicExpressionTree> SymbolicExpressionTrees {
+    public ItemArray<ISymbolicExpressionTree> SymbolicExpressionTree {
       get { return SymbolicExpressionTreeParameter.ActualValue; }
     }

trunk/sources/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Analyzers/SymbolicDataAnalysisMultiObjectiveTrainingBestSolutionAnalyzer.cs

@@ -87 +87 @@
       #region find best trees
       IList<int> nonDominatedIndexes = new List<int>();
-      ISymbolicExpressionTree[] tree = SymbolicExpressionTrees.ToArray();
+      ISymbolicExpressionTree[] tree = SymbolicExpressionTree.ToArray();
       List<double[]> qualities = Qualities.Select(x => x.ToArray()).ToList();
       bool[] maximization = Maximization.ToArray();

trunk/sources/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Analyzers/SymbolicDataAnalysisMultiObjectiveValidationAnalyzer.cs

@@ -20 +20 @@
 #endregion
 
+using System.Linq;
 using HeuristicLab.Common;
 using HeuristicLab.Core;
@@ -25 +26 @@
 using HeuristicLab.Parameters;
 using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
+using System.Collections.Generic;
+using System;
+using HeuristicLab.Random;
 
 namespace HeuristicLab.Problems.DataAnalysis.Symbolic {
@@ -35 +39 @@
     where T : class, ISymbolicDataAnalysisMultiObjectiveEvaluator<U>
     where U : class, IDataAnalysisProblemData {
+    private const string RandomParameterName = "Random";
     private const string ProblemDataParameterName = "ProblemData";
     private const string EvaluatorParameterName = "Evaluator";
@@ -42 +47 @@
 
     #region parameter properties
+    public ILookupParameter<IRandom> RandomParameter {
+      get { return (ILookupParameter<IRandom>)Parameters[RandomParameterName]; }
+    }
     public ILookupParameter<U> ProblemDataParameter {
       get { return (ILookupParameter<U>)Parameters[ProblemDataParameterName]; }
@@ -72 +80 @@
       Parameters.Add(new ValueLookupParameter<PercentValue>(RelativeNumberOfEvaluatedSamplesParameterName, "The relative number of samples of the dataset partition, which should be randomly chosen for evaluation between the start and end index."));
     }
+
+    protected IEnumerable<int> GenerateRowsToEvaluate() {
+      int seed = RandomParameter.ActualValue.Next();
+      int samplesStart = ValidationPartitionParameter.ActualValue.Start;
+      int samplesEnd = ValidationPartitionParameter.ActualValue.End;
+      int testPartitionStart = ProblemDataParameter.ActualValue.TestPartition.Start;
+      int testPartitionEnd = ProblemDataParameter.ActualValue.TestPartition.End;
+
+      if (samplesEnd < samplesStart) throw new ArgumentException("Start value is larger than end value.");
+      int count = (int)((samplesEnd - samplesStart) * RelativeNumberOfEvaluatedSamplesParameter.ActualValue.Value);
+      if (count == 0) count = 1;
+      return RandomEnumerable.SampleRandomNumbers(seed, samplesStart, samplesEnd, count)
+        .Where(i => i < testPartitionStart || testPartitionEnd <= i);
+    }
   }
 }

trunk/sources/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Analyzers/SymbolicDataAnalysisMultiObjectiveValidationBestSolutionAnalyzer.cs

@@ -75 +75 @@
 
     public override IOperation Apply() {
-      int start = ValidationPartitionParameter.ActualValue.Start;
-      int end = ValidationPartitionParameter.ActualValue.End;
-      int count = (int)((end - start) * RelativeNumberOfEvaluatedSamplesParameter.ActualValue.Value);
-      if (count <= 0) return base.Apply();
+      IEnumerable<int> rows = GenerateRowsToEvaluate();
+      if (rows.Count() <= 0) return base.Apply();
 
       var results = ResultCollection;
@@ -95 +93 @@
       #region find best trees
       IList<int> nonDominatedIndexes = new List<int>();
-      ISymbolicExpressionTree[] tree = SymbolicExpressionTrees.ToArray();
+      ISymbolicExpressionTree[] tree = SymbolicExpressionTree.ToArray();
       List<double[]> qualities = new List<double[]>();
       bool[] maximization = Maximization.ToArray();
       List<double[]> newNonDominatedQualities = new List<double[]>();
       var evaluator = EvaluatorParameter.ActualValue;
-      IEnumerable<int> rows = RandomEnumerable.SampleRandomNumbers(start, end, count);
       IExecutionContext childContext = (IExecutionContext)ExecutionContext.CreateChildOperation(evaluator);
       for (int i = 0; i < tree.Length; i++) {

trunk/sources/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Analyzers/SymbolicDataAnalysisSingleObjectiveTrainingBestSolutionAnalyzer.cs

@@ -74 +74 @@
       double bestQuality = Maximization.Value ? double.NegativeInfinity : double.PositiveInfinity;
       ISymbolicExpressionTree bestTree = null;
-      ISymbolicExpressionTree[] tree = SymbolicExpressionTrees.ToArray();
+      ISymbolicExpressionTree[] tree = SymbolicExpressionTree.ToArray();
       double[] quality = Quality.Select(x => x.Value).ToArray();
       for (int i = 0; i < tree.Length; i++) {

trunk/sources/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Analyzers/SymbolicDataAnalysisSingleObjectiveValidationAnalyzer.cs

@@ -20 +20 @@
 #endregion
 
+using System.Linq;
 using HeuristicLab.Common;
 using HeuristicLab.Core;
@@ -25 +26 @@
 using HeuristicLab.Parameters;
 using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
+using System.Collections.Generic;
+using System;
+using HeuristicLab.Random;
 
 namespace HeuristicLab.Problems.DataAnalysis.Symbolic {
@@ -35 +39 @@
     where T : class, ISymbolicDataAnalysisSingleObjectiveEvaluator<U>
     where U : class, IDataAnalysisProblemData {
+    private const string RandomParameterName = "Random";
     private const string ProblemDataParameterName = "ProblemData";
     private const string EvaluatorParameterName = "Evaluator";
@@ -42 +47 @@
 
     #region parameter properties
+    public ILookupParameter<IRandom> RandomParameter {
+      get { return (ILookupParameter<IRandom>)Parameters[RandomParameterName]; }
+    }
     public ILookupParameter<U> ProblemDataParameter {
       get { return (ILookupParameter<U>)Parameters[ProblemDataParameterName]; }
@@ -66 +74 @@
     public SymbolicDataAnalysisSingleObjectiveValidationAnalyzer()
       : base() {
+      Parameters.Add(new LookupParameter<IRandom>(RandomParameterName, "The random generator."));
       Parameters.Add(new LookupParameter<U>(ProblemDataParameterName, "The problem data of the symbolic data analysis problem."));
       Parameters.Add(new LookupParameter<T>(EvaluatorParameterName, "The operator to use for fitness evaluation on the validation partition."));
@@ -72 +81 @@
       Parameters.Add(new ValueLookupParameter<PercentValue>(RelativeNumberOfEvaluatedSamplesParameterName, "The relative number of samples of the dataset partition, which should be randomly chosen for evaluation between the start and end index."));
     }
+
+    protected IEnumerable<int> GenerateRowsToEvaluate() {
+      int seed = RandomParameter.ActualValue.Next();
+      int samplesStart = ValidationPartitionParameter.ActualValue.Start;
+      int samplesEnd = ValidationPartitionParameter.ActualValue.End;
+      int testPartitionStart = ProblemDataParameter.ActualValue.TestPartition.Start;
+      int testPartitionEnd = ProblemDataParameter.ActualValue.TestPartition.End;
+
+      if (samplesEnd < samplesStart) throw new ArgumentException("Start value is larger than end value.");
+      int count = (int)((samplesEnd - samplesStart) * RelativeNumberOfEvaluatedSamplesParameter.ActualValue.Value);
+      if (count == 0) count = 1;
+      return RandomEnumerable.SampleRandomNumbers(seed, samplesStart, samplesEnd, count)
+        .Where(i => i < testPartitionStart || testPartitionEnd <= i);
+    }
   }
 }

trunk/sources/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Analyzers/SymbolicDataAnalysisSingleObjectiveValidationBestSolutionAnalyzer.cs

@@ -76 +76 @@
       double bestQuality = Maximization.Value ? double.NegativeInfinity : double.PositiveInfinity;
       ISymbolicExpressionTree bestTree = null;
-      ISymbolicExpressionTree[] tree = SymbolicExpressionTrees.ToArray();
+      ISymbolicExpressionTree[] tree = SymbolicExpressionTree.ToArray();
       double[] quality = new double[tree.Length];
       var evaluator = EvaluatorParameter.ActualValue;
-      int start = ValidationPartitionParameter.ActualValue.Start;
-      int end = ValidationPartitionParameter.ActualValue.End;
-      int count = (int)((end - start) * RelativeNumberOfEvaluatedSamplesParameter.ActualValue.Value);
-      if (count <= 0) return base.Apply();
+      IEnumerable<int> rows = GenerateRowsToEvaluate();
+      if (rows.Count() <= 0) return base.Apply();
 
-      IEnumerable<int> rows = RandomEnumerable.SampleRandomNumbers(start, end, count);
       IExecutionContext childContext = (IExecutionContext)ExecutionContext.CreateChildOperation(evaluator);
       for (int i = 0; i < tree.Length; i++) {