Free cookie consent management tool by TermsFeed Policy Generator

Ignore:
Timestamp:
03/12/14 15:23:47 (11 years ago)
Author:
mkommend
Message:

#1997: Merged trunk changes into data analysis island algorithms branch and fixed bugs in the evaluators.

Location:
branches/DataAnalysis.IslandAlgorithms/HeuristicLab.Problems.DataAnalysis.Symbolic
Files:
12 edited
1 copied

Legend:

Unmodified
Added
Removed

  • branches/DataAnalysis.IslandAlgorithms/HeuristicLab.Problems.DataAnalysis.Symbolic

  • branches/DataAnalysis.IslandAlgorithms/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Analyzers/SymbolicDataAnalysisAlleleFrequencyAnalyzer.cs

    r9456 r10579  
    7575    public static Allele[] CalculateAlleles(ISymbolicExpressionTree solution, int alleleTreedepth) {
    7676      return GetAllSubtreesOfDepth(solution, alleleTreedepth)
    77         .AsParallel()
    7877        .Select(t => GetAlleleFromSubtreeOfDepth(t, alleleTreedepth))
    7978        .ToArray();

  • branches/DataAnalysis.IslandAlgorithms/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Analyzers/SymbolicDataAnalysisMultiObjectiveValidationBestSolutionAnalyzer.cs

    r9456 r10579  
    121121
    122122      var qualities = tree
    123         .AsParallel()
    124123        .Select(t => evaluator.Evaluate(childContext, t, problemData, rows))
    125124        .ToArray();

  • branches/DataAnalysis.IslandAlgorithms/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Analyzers/SymbolicDataAnalysisSingleObjectivePruningAnalyzer.cs

    r10390 r10579  
    55using HeuristicLab.Core;
    66using HeuristicLab.Data;
    7 using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
    8 using HeuristicLab.Optimization;
     7using HeuristicLab.Operators;
     8using HeuristicLab.Optimization.Operators;
    99using HeuristicLab.Parameters;
    1010using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
    … …  
    1414  [Item("SymbolicDataAnalysisSingleObjectivePruningAnalyzer", "An analyzer that prunes introns from trees in single objective symbolic data analysis problems.")]
    1515  public abstract class SymbolicDataAnalysisSingleObjectivePruningAnalyzer : SymbolicDataAnalysisSingleObjectiveAnalyzer {
     16    #region parameter names
    1617    private const string ProblemDataParameterName = "ProblemData";
    17     private const string InterpreterParameterName = "SymbolicExpressionTreeInterpreter";
    18 
    1918    private const string UpdateIntervalParameterName = "UpdateInverval";
    2019    private const string UpdateCounterParameterName = "UpdateCounter";
    21 
    2220    private const string PopulationSliceParameterName = "PopulationSlice";
    2321    private const string PruningProbabilityParameterName = "PruningProbability";
    24 
    25     private const string NumberOfPrunedSubtreesParameterName = "PrunedSubtrees";
    26     private const string NumberOfPrunedTreesParameterName = "PrunedTrees";
    27 
     22    private const string TotalNumberOfPrunedSubtreesParameterName = "Number of pruned subtrees";
     23    private const string TotalNumberOfPrunedTreesParameterName = "Number of pruned trees";
    2824    private const string RandomParameterName = "Random";
    29     private const string EstimationLimitsParameterName = "EstimationLimits";
    30 
    3125    private const string PruneOnlyZeroImpactNodesParameterName = "PruneOnlyZeroImpactNodes";
    3226    private const string NodeImpactThresholdParameterName = "ImpactThreshold";
    33 
    34     private bool reentry;
    35     [Storable]
    36     protected ISymbolicDataAnalysisSolutionImpactValuesCalculator impactValuesCalculator;
    37 
     27    private const string PruningOperatorParameterName = "PruningOperator";
     28    private const string ResultsParameterName = "Results";
     29    #endregion
     30    #region private members
     31    private DataReducer prunedSubtreesReducer;
     32    private DataReducer prunedTreesReducer;
     33    private DataTableValuesCollector valuesCollector;
     34    private ResultsCollector resultsCollector;
     35    private EmptyOperator emptyOp;
     36    #endregion
    3837    #region parameter properties
     38    public IValueParameter<SymbolicDataAnalysisExpressionPruningOperator> PruningOperatorParameter {
     39      get { return (IValueParameter<SymbolicDataAnalysisExpressionPruningOperator>)Parameters[PruningOperatorParameterName]; }
     40    }
    3941    public IFixedValueParameter<BoolValue> PruneOnlyZeroImpactNodesParameter {
    4042      get { return (IFixedValueParameter<BoolValue>)Parameters[PruneOnlyZeroImpactNodesParameterName]; }
    … …  
    4345      get { return (IFixedValueParameter<DoubleValue>)Parameters[NodeImpactThresholdParameterName]; }
    4446    }
    45     public ILookupParameter<DoubleLimit> EstimationLimitsParameter {
    46       get { return (ILookupParameter<DoubleLimit>)Parameters[EstimationLimitsParameterName]; }
    47     }
    4847    public ILookupParameter<IRandom> RandomParameter {
    4948      get { return (ILookupParameter<IRandom>)Parameters[RandomParameterName]; }
    … …  
    5150    private ILookupParameter<IDataAnalysisProblemData> ProblemDataParameter {
    5251      get { return (ILookupParameter<IDataAnalysisProblemData>)Parameters[ProblemDataParameterName]; }
    53     }
    54     private ILookupParameter<ISymbolicDataAnalysisExpressionTreeInterpreter> InterpreterParameter {
    55       get { return (ILookupParameter<ISymbolicDataAnalysisExpressionTreeInterpreter>)Parameters[InterpreterParameterName]; }
    5652    }
    5753    public IValueParameter<IntValue> UpdateIntervalParameter {
    … …  
    6763      get { return (IValueParameter<DoubleValue>)Parameters[PruningProbabilityParameterName]; }
    6864    }
    69     public IFixedValueParameter<DoubleValue> NumberOfPrunedSubtreesParameter {
    70       get { return (IFixedValueParameter<DoubleValue>)Parameters[NumberOfPrunedSubtreesParameterName]; }
    71     }
    72     public IFixedValueParameter<DoubleValue> NumberOfPrunedTreesParameter {
    73       get { return (IFixedValueParameter<DoubleValue>)Parameters[NumberOfPrunedTreesParameterName]; }
    74     }
    7565    #endregion
    7666    #region properties
     67    protected SymbolicDataAnalysisExpressionPruningOperator PruningOperator { get { return PruningOperatorParameter.Value; } }
    7768    protected IDataAnalysisProblemData ProblemData { get { return ProblemDataParameter.ActualValue; } }
    78     protected ISymbolicDataAnalysisExpressionTreeInterpreter Interpreter { get { return InterpreterParameter.ActualValue; } }
    7969    protected IntValue UpdateInterval { get { return UpdateIntervalParameter.Value; } }
    8070    protected IntValue UpdateCounter { get { return UpdateCounterParameter.Value; } }
    8171    protected DoubleRange PopulationSlice { get { return PopulationSliceParameter.Value; } }
    8272    protected DoubleValue PruningProbability { get { return PruningProbabilityParameter.Value; } }
    83     protected DoubleValue PrunedSubtrees { get { return NumberOfPrunedSubtreesParameter.Value; } }
    84     protected DoubleValue PrunedTrees { get { return NumberOfPrunedTreesParameter.Value; } }
    85     protected DoubleLimit EstimationLimits { get { return EstimationLimitsParameter.ActualValue; } }
    8673    protected IRandom Random { get { return RandomParameter.ActualValue; } }
    8774    protected DoubleValue NodeImpactThreshold { get { return NodeImpactThresholdParameter.Value; } }
    8875    protected BoolValue PruneOnlyZeroImpactNodes { get { return PruneOnlyZeroImpactNodesParameter.Value; } }
     76    #endregion
     77    #region IStatefulItem members
     78    public override void InitializeState() {
     79      base.InitializeState();
     80      UpdateCounter.Value = 0;
     81    }
     82    public override void ClearState() {
     83      base.ClearState();
     84      UpdateCounter.Value = 0;
     85    }
    8986    #endregion
    9087
    … …  
    9390    protected SymbolicDataAnalysisSingleObjectivePruningAnalyzer(SymbolicDataAnalysisSingleObjectivePruningAnalyzer original, Cloner cloner)
    9491      : base(original, cloner) {
    95       impactValuesCalculator = original.impactValuesCalculator;
     92      if (original.prunedSubtreesReducer != null)
     93        this.prunedSubtreesReducer = (DataReducer)original.prunedSubtreesReducer.Clone();
     94      if (original.prunedTreesReducer != null)
     95        this.prunedTreesReducer = (DataReducer)original.prunedTreesReducer.Clone();
     96      if (original.valuesCollector != null)
     97        this.valuesCollector = (DataTableValuesCollector)original.valuesCollector.Clone();
     98      if (original.resultsCollector != null)
     99        this.resultsCollector = (ResultsCollector)original.resultsCollector.Clone();
    96100    }
    97101    protected SymbolicDataAnalysisSingleObjectivePruningAnalyzer() {
     102      #region add parameters
    98103      Parameters.Add(new ValueParameter<DoubleRange>(PopulationSliceParameterName, new DoubleRange(0.75, 1)));
    99104      Parameters.Add(new ValueParameter<DoubleValue>(PruningProbabilityParameterName, new DoubleValue(0.5)));
    100       // analyzer parameters
    101105      Parameters.Add(new ValueParameter<IntValue>(UpdateIntervalParameterName, "The interval in which the tree length analysis should be applied.", new IntValue(1)));
    102106      Parameters.Add(new ValueParameter<IntValue>(UpdateCounterParameterName, "The value which counts how many times the operator was called", new IntValue(0)));
    103107      Parameters.Add(new LookupParameter<IRandom>(RandomParameterName));
    104108      Parameters.Add(new LookupParameter<IDataAnalysisProblemData>(ProblemDataParameterName));
    105       Parameters.Add(new LookupParameter<ISymbolicDataAnalysisExpressionTreeInterpreter>(InterpreterParameterName));
    106 
    107       Parameters.Add(new FixedValueParameter<DoubleValue>(NumberOfPrunedSubtreesParameterName, new DoubleValue(0)));
    108       Parameters.Add(new FixedValueParameter<DoubleValue>(NumberOfPrunedTreesParameterName, new DoubleValue(0)));
    109       Parameters.Add(new LookupParameter<DoubleLimit>(EstimationLimitsParameterName));
    110109      Parameters.Add(new FixedValueParameter<DoubleValue>(NodeImpactThresholdParameterName, new DoubleValue(0.0)));
    111110      Parameters.Add(new FixedValueParameter<BoolValue>(PruneOnlyZeroImpactNodesParameterName, new BoolValue(false)));
     111      #endregion
     112    }
     113
     114    private void InitializeOperators() {
     115      prunedSubtreesReducer = new DataReducer();
     116      prunedSubtreesReducer.ParameterToReduce.ActualName = PruningOperator.PrunedSubtreesParameter.ActualName;
     117      prunedSubtreesReducer.ReductionOperation.Value = new ReductionOperation(ReductionOperations.Sum); // sum all the pruned subtrees parameter values
     118      prunedSubtreesReducer.TargetOperation.Value = new ReductionOperation(ReductionOperations.Assign); // asign the sum to the target parameter
     119      prunedSubtreesReducer.TargetParameter.ActualName = TotalNumberOfPrunedSubtreesParameterName;
     120
     121      prunedTreesReducer = new DataReducer();
     122      prunedTreesReducer.ParameterToReduce.ActualName = PruningOperator.PrunedTreesParameter.ActualName;
     123      prunedTreesReducer.ReductionOperation.Value = new ReductionOperation(ReductionOperations.Sum);
     124      prunedTreesReducer.TargetOperation.Value = new ReductionOperation(ReductionOperations.Assign);
     125      prunedTreesReducer.TargetParameter.ActualName = TotalNumberOfPrunedTreesParameterName;
     126
     127      valuesCollector = new DataTableValuesCollector();
     128      valuesCollector.CollectedValues.Add(new LookupParameter<IntValue>(TotalNumberOfPrunedSubtreesParameterName));
     129      valuesCollector.CollectedValues.Add(new LookupParameter<IntValue>(TotalNumberOfPrunedTreesParameterName));
     130      valuesCollector.DataTableParameter.ActualName = "Population pruning";
     131
     132      resultsCollector = new ResultsCollector();
     133      resultsCollector.CollectedValues.Add(new LookupParameter<DataTable>("Population pruning"));
     134      resultsCollector.ResultsParameter.ActualName = ResultsParameterName;
     135
     136      emptyOp = new EmptyOperator();
     137    }
     138
     139    //
     140    /// <summary>
     141    /// Computes the closed interval bounding the portion of the population that is to be pruned.
     142    /// </summary>
     143    /// <returns>Returns an int range [start, end]</returns>
     144    private IntRange GetSliceBounds() {
     145      var count = ExecutionContext.Scope.SubScopes.Count;
     146      var start = (int)Math.Round(PopulationSlice.Start * count);
     147      var end = (int)Math.Round(PopulationSlice.End * count);
     148      if (end > count) end = count;
     149
     150      if (start >= end) throw new ArgumentOutOfRangeException("Invalid PopulationSlice bounds.");
     151      return new IntRange(start, end);
     152    }
     153
     154    private IOperation CreatePruningOperation() {
     155      var oc = new OperationCollection { Parallel = true };
     156      var range = GetSliceBounds();
     157      var qualities = Quality.Select(x => x.Value).ToArray();
     158      var indices = Enumerable.Range(0, qualities.Length).ToArray();
     159      Array.Sort(qualities, indices);
     160      if (!Maximization.Value) Array.Reverse(indices);
     161
     162      var subscopes = ExecutionContext.Scope.SubScopes;
     163
     164      for (int i = 0; i < subscopes.Count; ++i) {
     165        IOperator op;
     166        if (range.Start <= i && i < range.End && Random.NextDouble() <= PruningProbability.Value)
     167          op = PruningOperator;
     168        else op = emptyOp;
     169        var index = indices[i];
     170        var subscope = subscopes[index];
     171        oc.Add(ExecutionContext.CreateChildOperation(op, subscope));
     172      }
     173      return oc;
    112174    }
    113175
    114176    public override IOperation Apply() {
    115       if (reentry) {
    116         UpdateCounter.Value++;
     177      UpdateCounter.Value++;
     178      if (UpdateCounter.Value != UpdateInterval.Value) return base.Apply();
     179      UpdateCounter.Value = 0;
    117180
    118         if (UpdateCounter.Value != UpdateInterval.Value) return base.Apply();
    119         UpdateCounter.Value = 0;
     181      if (prunedSubtreesReducer == null || prunedTreesReducer == null || valuesCollector == null || resultsCollector == null) { InitializeOperators(); }
    120182
    121         var trees = SymbolicExpressionTreeParameter.ActualValue.ToList();
    122         var qualities = QualityParameter.ActualValue.ToList();
     183      var prune = CreatePruningOperation();
     184      var reducePrunedSubtrees = ExecutionContext.CreateChildOperation(prunedSubtreesReducer);
     185      var reducePrunedTrees = ExecutionContext.CreateChildOperation(prunedTreesReducer);
     186      var collectValues = ExecutionContext.CreateChildOperation(valuesCollector);
     187      var collectResults = ExecutionContext.CreateChildOperation(resultsCollector);
    123188
    124         var population = trees.Zip(qualities, (tree, quality) => new { Tree = tree, Quality = quality }).ToList();
    125         Func<double, double, int> compare = (a, b) => Maximization.Value ? a.CompareTo(b) : b.CompareTo(a);
    126         population.Sort((a, b) => compare(a.Quality.Value, b.Quality.Value));
    127 
    128         var start = (int)Math.Round(PopulationSlice.Start * trees.Count);
    129         var end = (int)Math.Round(PopulationSlice.End * trees.Count);
    130 
    131         if (end == population.Count) end--;
    132 
    133         if (start >= end || end >= population.Count) throw new Exception("Invalid PopulationSlice bounds.");
    134 
    135         PrunedSubtrees.Value = 0;
    136         PrunedTrees.Value = 0;
    137 
    138         reentry = false;
    139 
    140         var operations = new OperationCollection { Parallel = true };
    141         foreach (var p in population.Skip(start).Take(end)) {
    142           if (Random.NextDouble() > PruningProbability.Value) continue;
    143           var op = new SymbolicDataAnalysisExpressionPruningOperator {
    144             Model = CreateModel(p.Tree, Interpreter, EstimationLimits.Lower, EstimationLimits.Upper),
    145             ImpactsCalculator = impactValuesCalculator,
    146             ProblemData = ProblemData,
    147             Random = Random,
    148             PruneOnlyZeroImpactNodes = PruneOnlyZeroImpactNodes.Value,
    149             NodeImpactThreshold = NodeImpactThreshold.Value
    150           };
    151           operations.Add(ExecutionContext.CreateChildOperation(op, ExecutionContext.Scope));
    152         }
    153         return new OperationCollection { operations, ExecutionContext.CreateOperation(this) };
    154       }
    155 
    156       DataTable table;
    157 
    158       if (ResultCollection.ContainsKey("Population Pruning")) {
    159         table = (DataTable)ResultCollection["Population Pruning"].Value;
    160       } else {
    161         table = new DataTable("Population Pruning");
    162         table.Rows.Add(new DataRow("Pruned Trees") { VisualProperties = { StartIndexZero = true } });
    163         table.Rows.Add(new DataRow("Pruned Subtrees") { VisualProperties = { StartIndexZero = true } });
    164         ResultCollection.Add(new Result("Population Pruning", table));
    165       }
    166 
    167       table.Rows["Pruned Trees"].Values.Add(PrunedTrees.Value);
    168       table.Rows["Pruned Subtrees"].Values.Add(PrunedSubtrees.Value);
    169 
    170       reentry = true;
    171 
    172       return base.Apply();
     189      return new OperationCollection { prune, reducePrunedSubtrees, reducePrunedTrees, collectValues, collectResults, base.Apply() };
    173190    }
    174 
    175     protected abstract ISymbolicDataAnalysisModel CreateModel(ISymbolicExpressionTree tree, ISymbolicDataAnalysisExpressionTreeInterpreter interpreter,
    176       double lowerEstimationLimit = double.MinValue, double upperEstimationLimit = double.MaxValue);
    177191  }
    178192}

  • branches/DataAnalysis.IslandAlgorithms/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Analyzers/SymbolicDataAnalysisSingleObjectiveValidationBestSolutionAnalyzer.cs

    r9456 r10579  
    121121      var quality = tree
    122122        .Take(topN)
    123         .AsParallel()
    124123        .Select(t => evaluator.Evaluate(childContext, t, problemData, rows))
    125124        .ToArray();

  • branches/DataAnalysis.IslandAlgorithms/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Analyzers/SymbolicDataAnalysisSingleObjectiveValidationParetoBestSolutionAnalyzer.cs

    r9456 r10579  
    120120        .Select(i => tree[i])
    121121        .Take(topN)
    122         .AsParallel()
    123122        .Select(t => evaluator.Evaluate(childContext, t, problemData, rows))
    124123        .ToArray();

  • branches/DataAnalysis.IslandAlgorithms/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Analyzers/SymbolicDataAnalysisVariableFrequencyAnalyzer.cs

    r9456 r10579  
    131131
    132132      var variableFrequencies = trees
    133         .AsParallel()
    134133        .SelectMany(t => GetVariableReferences(t, aggregateLaggedVariables))
    135134        .GroupBy(pair => pair.Key, pair => pair.Value)

  • branches/DataAnalysis.IslandAlgorithms/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/HeuristicLab.Problems.DataAnalysis.Symbolic-3.4.csproj

    r10390 r10579  
    139139      <Private>False</Private>
    140140    </Reference>
     141    <Reference Include="HeuristicLab.Optimization.Operators-3.3, Version=3.3.0.0, Culture=neutral, PublicKeyToken=ba48961d6f65dcec, processorArchitecture=MSIL">
     142      <SpecificVersion>False</SpecificVersion>
     143      <HintPath>..\..\..\..\trunk\sources\bin\HeuristicLab.Optimization.Operators-3.3.dll</HintPath>
     144      <Private>False</Private>
     145    </Reference>
    141146    <Reference Include="HeuristicLab.Parameters-3.3">
    142147      <HintPath>..\..\..\..\trunk\sources\bin\HeuristicLab.Parameters-3.3.dll</HintPath>
    … …  
    192197      <SubType>Code</SubType>
    193198    </Compile>
     199    <Compile Include="Matching\SymbolicExpressionTreeCanonicalSorter.cs" />
     200    <Compile Include="Matching\SymbolicExpressionTreeEqualityComparer.cs" />
     201    <Compile Include="Matching\SymbolicExpressionTreeMatching.cs" />
     202    <Compile Include="Matching\SymbolicExpressionTreeMaxCommonSequenceCalculator.cs" />
     203    <Compile Include="Matching\SymbolicExpressionTreeNodeComparer.cs" />
     204    <Compile Include="Matching\SymbolicExpressionTreeNodeSimilarityComparer.cs" />
    194205    <Compile Include="SymbolicDataAnalysisExpressionPruningOperator.cs" />
    195206    <Compile Include="Analyzers\SymbolicDataAnalysisVariableFrequencyAnalyzer.cs" />

  • branches/DataAnalysis.IslandAlgorithms/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Interfaces/ISymbolicDataAnalysisImpactValuesCalculator.cs

    r8946 r10579  
    11using System.Collections.Generic;
     2using HeuristicLab.Core;
    23using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
    34
    45namespace HeuristicLab.Problems.DataAnalysis.Symbolic {
    5   public interface ISymbolicDataAnalysisSolutionImpactValuesCalculator {
     6  public interface ISymbolicDataAnalysisSolutionImpactValuesCalculator : IItem {
    67    double CalculateReplacementValue(ISymbolicDataAnalysisModel model, ISymbolicExpressionTreeNode node, IDataAnalysisProblemData problemData, IEnumerable<int> rows);
    78    double CalculateImpactValue(ISymbolicDataAnalysisModel model, ISymbolicExpressionTreeNode node, IDataAnalysisProblemData problemData, IEnumerable<int> rows, double originalQuality = double.NaN);
     9    void CalculateImpactAndReplacementValues(ISymbolicDataAnalysisModel model, ISymbolicExpressionTreeNode node, IDataAnalysisProblemData problemData,
     10      IEnumerable<int> rows, out double impactValue, out double replacementValue, double originalQuality = double.NaN);
    811  }
    912}

  • branches/DataAnalysis.IslandAlgorithms/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Plugin.cs.frame

    r10037 r10579  
    3838  [PluginDependency("HeuristicLab.Operators", "3.3")]
    3939  [PluginDependency("HeuristicLab.Optimization", "3.3")]
     40  [PluginDependency("HeuristicLab.Optimization.Operators", "3.3")]
    4041  [PluginDependency("HeuristicLab.Parameters", "3.3")]
    4142  [PluginDependency("HeuristicLab.Persistence", "3.3")]

  • branches/DataAnalysis.IslandAlgorithms/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/SymbolicDataAnalysisExpressionPruningOperator.cs

    r10390 r10579  
    3434  [StorableClass]
    3535  [Item("SymbolicExpressionTreePruningOperator", "An operator that replaces introns with constant values in a symbolic expression tree.")]
    36   public class SymbolicDataAnalysisExpressionPruningOperator : SingleSuccessorOperator {
    37     private const string NumberOfPrunedSubtreesParameterName = "PrunedSubtrees";
    38     private const string NumberOfPrunedTreesParameterName = "PrunedTrees";
     36  public abstract class SymbolicDataAnalysisExpressionPruningOperator : SingleSuccessorOperator {
     37    #region parameter names
     38    private const string ProblemDataParameterName = "ProblemData";
     39    private const string SymbolicDataAnalysisModelParameterName = "SymbolicDataAnalysisModel";
     40    private const string ImpactValuesCalculatorParameterName = "ImpactValuesCalculator";
     41    private const string PrunedSubtreesParameterName = "PrunedSubtrees";
     42    private const string PrunedTreesParameterName = "PrunedTrees";
     43    private const string FitnessCalculationPartitionParameterName = "FitnessCalculationPartition";
     44    private const string NodeImpactThresholdParameterName = "ImpactThreshold";
     45    private const string PruneOnlyZeroImpactNodesParameterName = "PruneOnlyZeroImpactNodes";
     46    private const string SymbolicExpressionTreeParameterName = "SymbolicExpressionTree"; // the tree to be pruned
     47    private const string QualityParameterName = "Quality"; // the quality
     48    private const string EstimationLimitsParameterName = "EstimationLimits";
     49    private const string InterpreterParameterName = "SymbolicExpressionTreeInterpreter";
     50    #endregion
     51
    3952    #region parameter properties
    40     public ILookupParameter<DoubleValue> NumberOfPrunedSubtreesParameter {
    41       get { return (ILookupParameter<DoubleValue>)Parameters[NumberOfPrunedSubtreesParameterName]; }
     53    public ILookupParameter<ISymbolicExpressionTree> SymbolicExpressionTreeParameter {
     54      get { return (ILookupParameter<ISymbolicExpressionTree>)Parameters[SymbolicExpressionTreeParameterName]; }
    4255    }
    43     public ILookupParameter<DoubleValue> NumberOfPrunedTreesParameter {
    44       get { return (ILookupParameter<DoubleValue>)Parameters[NumberOfPrunedTreesParameterName]; }
     56    public ILookupParameter<DoubleValue> QualityParameter {
     57      get { return (ILookupParameter<DoubleValue>)Parameters[QualityParameterName]; }
     58    }
     59    public ILookupParameter<IDataAnalysisProblemData> ProblemDataParameter {
     60      get { return (ILookupParameter<IDataAnalysisProblemData>)Parameters[ProblemDataParameterName]; }
     61    }
     62    public IValueParameter<ISymbolicDataAnalysisSolutionImpactValuesCalculator> ImpactValuesCalculatorParameter {
     63      get { return (IValueParameter<ISymbolicDataAnalysisSolutionImpactValuesCalculator>)Parameters[ImpactValuesCalculatorParameterName]; }
     64    }
     65    public ILookupParameter<IntRange> FitnessCalculationPartitionParameter {
     66      get { return (ILookupParameter<IntRange>)Parameters[FitnessCalculationPartitionParameterName]; }
     67    }
     68    public ILookupParameter<IntValue> PrunedSubtreesParameter {
     69      get { return (ILookupParameter<IntValue>)Parameters[PrunedSubtreesParameterName]; }
     70    }
     71    public ILookupParameter<IntValue> PrunedTreesParameter {
     72      get { return (ILookupParameter<IntValue>)Parameters[PrunedTreesParameterName]; }
     73    }
     74    public IFixedValueParameter<DoubleValue> NodeImpactThresholdParameter {
     75      get { return (IFixedValueParameter<DoubleValue>)Parameters[NodeImpactThresholdParameterName]; }
     76    }
     77    public IFixedValueParameter<BoolValue> PruneOnlyZeroImpactNodesParameter {
     78      get { return (IFixedValueParameter<BoolValue>)Parameters[PruneOnlyZeroImpactNodesParameterName]; }
     79    }
     80    public ILookupParameter<DoubleLimit> EstimationLimitsParameter {
     81      get { return (ILookupParameter<DoubleLimit>)Parameters[EstimationLimitsParameterName]; }
     82    }
     83    public ILookupParameter<ISymbolicDataAnalysisExpressionTreeInterpreter> InterpreterParameter {
     84      get { return (ILookupParameter<ISymbolicDataAnalysisExpressionTreeInterpreter>)Parameters[InterpreterParameterName]; }
    4585    }
    4686    #endregion
    4787    #region properties
    48     private DoubleValue PrunedSubtrees { get { return NumberOfPrunedSubtreesParameter.ActualValue; } }
    49     private DoubleValue PrunedTrees { get { return NumberOfPrunedTreesParameter.ActualValue; } }
     88    protected IDataAnalysisProblemData ProblemData { get { return ProblemDataParameter.ActualValue; } }
     89    protected ISymbolicDataAnalysisSolutionImpactValuesCalculator ImpactValuesCalculator { get { return ImpactValuesCalculatorParameter.Value; } }
     90    protected IntRange FitnessCalculationPartition { get { return FitnessCalculationPartitionParameter.ActualValue; } }
     91    protected BoolValue PruneOnlyZeroImpactNodes { get { return PruneOnlyZeroImpactNodesParameter.Value; } }
     92    protected DoubleValue NodeImpactThreshold { get { return NodeImpactThresholdParameter.Value; } }
     93    protected ISymbolicExpressionTree SymbolicExpressionTree { get { return SymbolicExpressionTreeParameter.ActualValue; } }
     94    protected DoubleValue Quality { get { return QualityParameter.ActualValue; } }
     95    protected DoubleLimit EstimationLimits { get { return EstimationLimitsParameter.ActualValue; } }
     96    protected ISymbolicDataAnalysisExpressionTreeInterpreter Interpreter { get { return InterpreterParameter.ActualValue; } }
    5097    #endregion
    5198
    5299    [StorableConstructor]
    53100    protected SymbolicDataAnalysisExpressionPruningOperator(bool deserializing) : base(deserializing) { }
    54     public override IDeepCloneable Clone(Cloner cloner) {
    55       return new SymbolicDataAnalysisExpressionPruningOperator(this, cloner);
     101    protected SymbolicDataAnalysisExpressionPruningOperator(SymbolicDataAnalysisExpressionPruningOperator original, Cloner cloner)
     102      : base(original, cloner) { }
     103
     104    protected SymbolicDataAnalysisExpressionPruningOperator() {
     105      #region add parameters
     106      Parameters.Add(new LookupParameter<IDataAnalysisProblemData>(ProblemDataParameterName));
     107      Parameters.Add(new LookupParameter<ISymbolicDataAnalysisModel>(SymbolicDataAnalysisModelParameterName));
     108      Parameters.Add(new LookupParameter<IntRange>(FitnessCalculationPartitionParameterName));
     109      Parameters.Add(new LookupParameter<IntValue>(PrunedSubtreesParameterName, "A counter of how many subtrees were replaced."));
     110      Parameters.Add(new LookupParameter<IntValue>(PrunedTreesParameterName, "A counter of how many trees were pruned."));
     111      Parameters.Add(new FixedValueParameter<BoolValue>(PruneOnlyZeroImpactNodesParameterName, "Specify whether or not only zero impact nodes should be pruned."));
     112      Parameters.Add(new FixedValueParameter<DoubleValue>(NodeImpactThresholdParameterName, "Specifies an impact value threshold below which nodes should be pruned."));
     113      Parameters.Add(new LookupParameter<DoubleLimit>(EstimationLimitsParameterName));
     114      Parameters.Add(new LookupParameter<ISymbolicDataAnalysisExpressionTreeInterpreter>(InterpreterParameterName));
     115      Parameters.Add(new LookupParameter<ISymbolicExpressionTree>(SymbolicExpressionTreeParameterName));
     116      Parameters.Add(new LookupParameter<DoubleValue>(QualityParameterName));
     117      #endregion
    56118    }
    57     protected SymbolicDataAnalysisExpressionPruningOperator(SymbolicDataAnalysisExpressionPruningOperator original, Cloner cloner)
    58       : base(original, cloner) {
    59     }
     119    public override IOperation Apply() {
     120      var model = CreateModel();
     121      var nodes = SymbolicExpressionTree.Root.GetSubtree(0).GetSubtree(0).IterateNodesPrefix().ToList();
     122      var rows = Enumerable.Range(FitnessCalculationPartition.Start, FitnessCalculationPartition.Size).ToList();
    60123
    61     public SymbolicDataAnalysisExpressionPruningOperator() {
    62       Parameters.Add(new LookupParameter<DoubleValue>(NumberOfPrunedSubtreesParameterName));
    63       Parameters.Add(new LookupParameter<DoubleValue>(NumberOfPrunedTreesParameterName));
    64     }
     124      var prunedSubtrees = 0;
     125      var prunedTrees = 0;
    65126
    66     public ISymbolicDataAnalysisModel Model { get; set; }
    67     public IDataAnalysisProblemData ProblemData { get; set; }
    68     public ISymbolicDataAnalysisSolutionImpactValuesCalculator ImpactsCalculator { get; set; }
    69     public IRandom Random { get; set; }
     127      double quality = Evaluate(model);
    70128
    71     public bool PruneOnlyZeroImpactNodes { get; set; }
    72     public double NodeImpactThreshold { get; set; }
    73 
    74     public override IOperation Apply() {
    75       int prunedSubtrees = 0;
    76 
    77       var nodes = Model.SymbolicExpressionTree.Root.GetSubtree(0).GetSubtree(0).IterateNodesPrefix().ToList();
    78 
    79       for (int j = 0; j < nodes.Count; ++j) {
    80         var node = nodes[j];
     129      for (int i = 0; i < nodes.Count; ++i) {
     130        var node = nodes[i];
    81131        if (node is ConstantTreeNode) continue;
    82132
    83         var impact = ImpactsCalculator.CalculateImpactValue(Model, node, ProblemData, ProblemData.TrainingIndices);
     133        double impactValue, replacementValue;
     134        ImpactValuesCalculator.CalculateImpactAndReplacementValues(model, node, ProblemData, rows, out impactValue, out replacementValue, quality);
    84135
    85         if (PruneOnlyZeroImpactNodes) {
    86           if (!impact.IsAlmost(0.0)) continue;
    87         } else {
    88           if (NodeImpactThreshold < impact) continue;
    89         }
     136        if (PruneOnlyZeroImpactNodes.Value && (!impactValue.IsAlmost(0.0))) continue;
     137        else if (NodeImpactThreshold.Value < impactValue) continue;
    90138
    91         var replacementValue = ImpactsCalculator.CalculateReplacementValue(Model, node, ProblemData, ProblemData.TrainingIndices);
    92139        var constantNode = new ConstantTreeNode(new Constant()) { Value = replacementValue };
    93140        ReplaceWithConstant(node, constantNode);
    94         j += node.GetLength() - 1; // skip subtrees under the node that was folded
     141        i += node.GetLength() - 1; // skip subtrees under the node that was folded
     142
     143        quality -= impactValue;
    95144
    96145        prunedSubtrees++;
    97146      }
    98147
    99       if (prunedSubtrees > 0) {
    100         lock (PrunedSubtrees) { PrunedSubtrees.Value += prunedSubtrees; }
    101         lock (PrunedTrees) { PrunedTrees.Value += 1; }
    102       }
     148      if (prunedSubtrees > 0) prunedTrees = 1;
     149      PrunedSubtreesParameter.ActualValue = new IntValue(prunedSubtrees);
     150      PrunedTreesParameter.ActualValue = new IntValue(prunedTrees);
     151
    103152      return base.Apply();
    104153    }
    … …  
    109158      parent.InsertSubtree(i, replacement);
    110159    }
     160    protected abstract ISymbolicDataAnalysisModel CreateModel();
     161    protected abstract double Evaluate(IDataAnalysisModel model);
    111162  }
    112163}

  • branches/DataAnalysis.IslandAlgorithms/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/SymbolicDataAnalysisSolutionImpactValuesCalculator.cs

    r9456 r10579  
    2222using System.Collections.Generic;
    2323using HeuristicLab.Common;
     24using HeuristicLab.Core;
    2425using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
     26using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
    2527
    2628namespace HeuristicLab.Problems.DataAnalysis.Symbolic {
    27   public abstract class SymbolicDataAnalysisSolutionImpactValuesCalculator : ISymbolicDataAnalysisSolutionImpactValuesCalculator {
     29  [StorableClass]
     30  [Item("SymbolicDataAnalysisSolutionImpactValuesCalculator", "Calculates the impact values and replacements values for symbolic expression tree nodes.")]
     31  public abstract class SymbolicDataAnalysisSolutionImpactValuesCalculator : Item, ISymbolicDataAnalysisSolutionImpactValuesCalculator {
     32    protected SymbolicDataAnalysisSolutionImpactValuesCalculator() { }
     33
     34    protected SymbolicDataAnalysisSolutionImpactValuesCalculator(SymbolicDataAnalysisSolutionImpactValuesCalculator original, Cloner cloner)
     35      : base(original, cloner) { }
     36    [StorableConstructor]
     37    protected SymbolicDataAnalysisSolutionImpactValuesCalculator(bool deserializing) : base(deserializing) { }
    2838    public abstract double CalculateReplacementValue(ISymbolicDataAnalysisModel model, ISymbolicExpressionTreeNode node, IDataAnalysisProblemData problemData, IEnumerable<int> rows);
    2939    public abstract double CalculateImpactValue(ISymbolicDataAnalysisModel model, ISymbolicExpressionTreeNode node, IDataAnalysisProblemData problemData, IEnumerable<int> rows, double originalQuality = double.NaN);
     40    public abstract void CalculateImpactAndReplacementValues(ISymbolicDataAnalysisModel model, ISymbolicExpressionTreeNode node, IDataAnalysisProblemData problemData, IEnumerable<int> rows, out double impactValue, out double replacementValue, double originalQuality = double.NaN);
    3041
    3142    protected static double CalculateReplacementValue(ISymbolicExpressionTreeNode node, ISymbolicExpressionTree sourceTree, ISymbolicDataAnalysisExpressionTreeInterpreter interpreter,
Note: See TracChangeset for help on using the changeset viewer.