Changeset 12745 for stable/HeuristicLab.Problems.DataAnalysis.Symbolic.Classification

Timestamp:

07/12/15 11:23:06 (9 years ago)

Author:

gkronber

Message:

#2359, #2398: merged r12189,r12358,r12359,r12361,r12461,r12674,r12720,r12744 from trunk to stable

Location:

stable

Files:

• . (modified) (1 prop)
• HeuristicLab.Problems.DataAnalysis.Symbolic.Classification (modified) (1 prop)
• HeuristicLab.Problems.DataAnalysis.Symbolic.Classification/3.4/SymbolicClassificationPruningAnalyzer.cs (modified) (3 diffs)
• HeuristicLab.Problems.DataAnalysis.Symbolic.Classification/3.4/SymbolicClassificationPruningOperator.cs (modified) (3 diffs)
• HeuristicLab.Problems.DataAnalysis.Symbolic.Classification/3.4/SymbolicClassificationSolutionImpactValuesCalculator.cs (modified) (2 diffs)

stable/HeuristicLab.Problems.DataAnalysis.Symbolic.Classification/3.4/SymbolicClassificationPruningAnalyzer.cs

@@ -22 +22 @@
 using HeuristicLab.Common;
 using HeuristicLab.Core;
+using HeuristicLab.Data;
 using HeuristicLab.Parameters;
 using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
@@ -29 +30 @@
   [StorableClass]
   public sealed class SymbolicClassificationPruningAnalyzer : SymbolicDataAnalysisSingleObjectivePruningAnalyzer {
-    private const string ImpactValuesCalculatorParameterName = "ImpactValuesCalculator";
     private const string PruningOperatorParameterName = "PruningOperator";
-    private SymbolicClassificationPruningAnalyzer(SymbolicClassificationPruningAnalyzer original, Cloner cloner)
-      : base(original, cloner) {
+    public IValueParameter<SymbolicClassificationPruningOperator> PruningOperatorParameter {
+      get { return (IValueParameter<SymbolicClassificationPruningOperator>)Parameters[PruningOperatorParameterName]; }
     }
-    public override IDeepCloneable Clone(Cloner cloner) {
-      return new SymbolicClassificationPruningAnalyzer(this, cloner);
+
+    protected override SymbolicDataAnalysisExpressionPruningOperator PruningOperator {
+      get { return PruningOperatorParameter.Value; }
     }
+
+    private SymbolicClassificationPruningAnalyzer(SymbolicClassificationPruningAnalyzer original, Cloner cloner) : base(original, cloner) { }
+    public override IDeepCloneable Clone(Cloner cloner) { return new SymbolicClassificationPruningAnalyzer(this, cloner); }
 
     [StorableConstructor]
@@ -42 +46 @@
 
     public SymbolicClassificationPruningAnalyzer() {
-      Parameters.Add(new ValueParameter<SymbolicDataAnalysisSolutionImpactValuesCalculator>(ImpactValuesCalculatorParameterName, "The impact values calculator", new SymbolicClassificationSolutionImpactValuesCalculator()));
-      Parameters.Add(new ValueParameter<SymbolicDataAnalysisExpressionPruningOperator>(PruningOperatorParameterName, "The operator used to prune trees", new SymbolicClassificationPruningOperator()));
+      Parameters.Add(new ValueParameter<SymbolicClassificationPruningOperator>(PruningOperatorParameterName, "The operator used to prune trees", new SymbolicClassificationPruningOperator(new SymbolicClassificationSolutionImpactValuesCalculator())));
+    }
+
+    [StorableHook(HookType.AfterDeserialization)]
+    private void AfterDeserialization() {
+      // BackwardsCompatibility3.3
+
+      #region Backwards compatible code, remove with 3.4
+      if (Parameters.ContainsKey(PruningOperatorParameterName)) {
+        var oldParam = Parameters[PruningOperatorParameterName] as ValueParameter<SymbolicDataAnalysisExpressionPruningOperator>;
+        if (oldParam != null) {
+          Parameters.Remove(oldParam);
+          Parameters.Add(new ValueParameter<SymbolicClassificationPruningOperator>(PruningOperatorParameterName, "The operator used to prune trees", new SymbolicClassificationPruningOperator(new SymbolicClassificationSolutionImpactValuesCalculator())));
+        }
+      } else {
+        // not yet contained
+        Parameters.Add(new ValueParameter<SymbolicClassificationPruningOperator>(PruningOperatorParameterName, "The operator used to prune trees", new SymbolicClassificationPruningOperator(new SymbolicClassificationSolutionImpactValuesCalculator())));
+      }
+
+      if (Parameters.ContainsKey("PruneOnlyZeroImpactNodes")) {
+        PruningOperator.PruneOnlyZeroImpactNodes = ((IFixedValueParameter<BoolValue>)Parameters["PruneOnlyZeroImpactNodes"]).Value.Value;
+        Parameters.Remove(Parameters["PruneOnlyZeroImpactNodes"]);
+      }
+      if (Parameters.ContainsKey("ImpactThreshold")) {
+        PruningOperator.NodeImpactThreshold = ((IFixedValueParameter<DoubleValue>)Parameters["ImpactThreshold"]).Value.Value;
+        Parameters.Remove(Parameters["ImpactThreshold"]);
+      }
+      if (Parameters.ContainsKey("ImpactValuesCalculator")) {
+        PruningOperator.ImpactValuesCalculator = ((ValueParameter<SymbolicDataAnalysisSolutionImpactValuesCalculator>)Parameters["ImpactValuesCalculator"]).Value;
+        Parameters.Remove(Parameters["ImpactValuesCalculator"]);
+      }
+      #endregion
     }
   }

stable/HeuristicLab.Problems.DataAnalysis.Symbolic.Classification/3.4/SymbolicClassificationPruningOperator.cs

@@ -22 +22 @@
 #endregion
 
+using System.Collections.Generic;
 using System.Linq;
 using HeuristicLab.Common;
 using HeuristicLab.Core;
+using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
 using HeuristicLab.Parameters;
 using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
@@ -32 +34 @@
   [Item("SymbolicClassificationPruningOperator", "An operator which prunes symbolic classificaton trees.")]
   public class SymbolicClassificationPruningOperator : SymbolicDataAnalysisExpressionPruningOperator {
-    private const string ImpactValuesCalculatorParameterName = "ImpactValuesCalculator";
     private const string ModelCreatorParameterName = "ModelCreator";
+    private const string EvaluatorParameterName = "Evaluator";
 
     #region parameter properties
@@ -39 +41 @@
       get { return (ILookupParameter<ISymbolicClassificationModelCreator>)Parameters[ModelCreatorParameterName]; }
     }
+
+    public ILookupParameter<ISymbolicClassificationSingleObjectiveEvaluator> EvaluatorParameter {
+      get {
+        return (ILookupParameter<ISymbolicClassificationSingleObjectiveEvaluator>)Parameters[EvaluatorParameterName];
+      }
+    }
     #endregion
 
-    protected SymbolicClassificationPruningOperator(SymbolicClassificationPruningOperator original, Cloner cloner)
-      : base(original, cloner) {
-    }
-
-    public override IDeepCloneable Clone(Cloner cloner) {
-      return new SymbolicClassificationPruningOperator(this, cloner);
-    }
+    protected SymbolicClassificationPruningOperator(SymbolicClassificationPruningOperator original, Cloner cloner) : base(original, cloner) { }
+    public override IDeepCloneable Clone(Cloner cloner) { return new SymbolicClassificationPruningOperator(this, cloner); }
 
     [StorableConstructor]
     protected SymbolicClassificationPruningOperator(bool deserializing) : base(deserializing) { }
 
-    public SymbolicClassificationPruningOperator() {
-      Parameters.Add(new ValueParameter<ISymbolicDataAnalysisSolutionImpactValuesCalculator>(ImpactValuesCalculatorParameterName, new SymbolicClassificationSolutionImpactValuesCalculator()));
+    public SymbolicClassificationPruningOperator(ISymbolicDataAnalysisSolutionImpactValuesCalculator impactValuesCalculator)
+      : base(impactValuesCalculator) {
       Parameters.Add(new LookupParameter<ISymbolicClassificationModelCreator>(ModelCreatorParameterName));
+      Parameters.Add(new LookupParameter<ISymbolicClassificationSingleObjectiveEvaluator>(EvaluatorParameterName));
     }
 
-    protected override ISymbolicDataAnalysisModel CreateModel() {
-      var model = ModelCreatorParameter.ActualValue.CreateSymbolicClassificationModel(SymbolicExpressionTree, Interpreter, EstimationLimits.Lower, EstimationLimits.Upper);
-      var problemData = (IClassificationProblemData)ProblemData;
-      var rows = problemData.TrainingIndices;
-      model.RecalculateModelParameters(problemData, rows);
+    [StorableHook(HookType.AfterDeserialization)]
+    private void AfterDeserialization() {
+      // BackwardsCompatibility3.3
+      #region Backwards compatible code, remove with 3.4
+      base.ImpactValuesCalculator = new SymbolicClassificationSolutionImpactValuesCalculator();
+      if (!Parameters.ContainsKey(EvaluatorParameterName)) {
+        Parameters.Add(new LookupParameter<ISymbolicClassificationSingleObjectiveEvaluator>(EvaluatorParameterName));
+      }
+      #endregion
+    }
+
+    protected override ISymbolicDataAnalysisModel CreateModel(ISymbolicExpressionTree tree, ISymbolicDataAnalysisExpressionTreeInterpreter interpreter, IDataAnalysisProblemData problemData, DoubleLimit estimationLimits) {
+      var model = ModelCreatorParameter.ActualValue.CreateSymbolicClassificationModel(tree, interpreter, estimationLimits.Lower, estimationLimits.Upper);
+      var classificationProblemData = (IClassificationProblemData)problemData;
+      var rows = classificationProblemData.TrainingIndices;
+      model.RecalculateModelParameters(classificationProblemData, rows);
       return model;
     }
 
     protected override double Evaluate(IDataAnalysisModel model) {
-      var classificationModel = (IClassificationModel)model;
-      var classificationProblemData = (IClassificationProblemData)ProblemData;
-      var trainingIndices = Enumerable.Range(FitnessCalculationPartition.Start, FitnessCalculationPartition.Size);
-      var estimatedValues = classificationModel.GetEstimatedClassValues(ProblemData.Dataset, trainingIndices);
-      var targetValues = ProblemData.Dataset.GetDoubleValues(classificationProblemData.TargetVariable, trainingIndices);
-      OnlineCalculatorError errorState;
-      var quality = OnlineAccuracyCalculator.Calculate(targetValues, estimatedValues, out errorState);
-      if (errorState != OnlineCalculatorError.None) return double.NaN;
-      return quality;
+      var evaluator = EvaluatorParameter.ActualValue;
+      var classificationModel = (ISymbolicClassificationModel)model;
+      var classificationProblemData = (IClassificationProblemData)ProblemDataParameter.ActualValue;
+      var rows = Enumerable.Range(FitnessCalculationPartitionParameter.ActualValue.Start, FitnessCalculationPartitionParameter.ActualValue.Size);
+      return evaluator.Evaluate(this.ExecutionContext, classificationModel.SymbolicExpressionTree, classificationProblemData, rows);
+    }
+
+    public static ISymbolicExpressionTree Prune(ISymbolicExpressionTree tree, ISymbolicClassificationModelCreator modelCreator,
+      SymbolicClassificationSolutionImpactValuesCalculator impactValuesCalculator, ISymbolicDataAnalysisExpressionTreeInterpreter interpreter,
+      IClassificationProblemData problemData, DoubleLimit estimationLimits, IEnumerable<int> rows,
+      double nodeImpactThreshold = 0.0, bool pruneOnlyZeroImpactNodes = false) {
+      var clonedTree = (ISymbolicExpressionTree)tree.Clone();
+      var model = modelCreator.CreateSymbolicClassificationModel(clonedTree, interpreter, estimationLimits.Lower, estimationLimits.Upper);
+
+      var nodes = clonedTree.Root.GetSubtree(0).GetSubtree(0).IterateNodesPrefix().ToList();
+      double qualityForImpactsCalculation = double.NaN;
+
+      for (int i = 0; i < nodes.Count; ++i) {
+        var node = nodes[i];
+        if (node is ConstantTreeNode) continue;
+
+        double impactValue, replacementValue, newQualityForImpactsCalculation;
+        impactValuesCalculator.CalculateImpactAndReplacementValues(model, node, problemData, rows, out impactValue, out replacementValue, out newQualityForImpactsCalculation, qualityForImpactsCalculation);
+
+        if (pruneOnlyZeroImpactNodes && !impactValue.IsAlmost(0.0)) continue;
+        if (!pruneOnlyZeroImpactNodes && impactValue > nodeImpactThreshold) continue;
+
+        var constantNode = (ConstantTreeNode)node.Grammar.GetSymbol("Constant").CreateTreeNode();
+        constantNode.Value = replacementValue;
+
+        ReplaceWithConstant(node, constantNode);
+        i += node.GetLength() - 1; // skip subtrees under the node that was folded
+
+        qualityForImpactsCalculation = newQualityForImpactsCalculation;
+      }
+      return model.SymbolicExpressionTree;
     }
   }

stable/HeuristicLab.Problems.DataAnalysis.Symbolic.Classification/3.4/SymbolicClassificationSolutionImpactValuesCalculator.cs

@@ -47 +47 @@
     }
 
-    public override double CalculateImpactValue(ISymbolicDataAnalysisModel model, ISymbolicExpressionTreeNode node, IDataAnalysisProblemData problemData, IEnumerable<int> rows, double originalQuality = double.NaN) {
+    public override double CalculateImpactValue(ISymbolicDataAnalysisModel model, ISymbolicExpressionTreeNode node, IDataAnalysisProblemData problemData, IEnumerable<int> rows, double qualityForImpactsCalculation = double.NaN) {
       double impactValue, replacementValue;
-      CalculateImpactAndReplacementValues(model, node, problemData, rows, out impactValue, out replacementValue, originalQuality);
+      double newQualityForImpactsCalculation;
+      CalculateImpactAndReplacementValues(model, node, problemData, rows, out impactValue, out replacementValue, out newQualityForImpactsCalculation, qualityForImpactsCalculation);
       return impactValue;
     }
 
     public override void CalculateImpactAndReplacementValues(ISymbolicDataAnalysisModel model, ISymbolicExpressionTreeNode node,
-      IDataAnalysisProblemData problemData, IEnumerable<int> rows, out double impactValue, out double replacementValue,
-      double originalQuality = Double.NaN) {
+      IDataAnalysisProblemData problemData, IEnumerable<int> rows, out double impactValue, out double replacementValue, out double newQualityForImpactsCalculation,
+      double qualityForImpactsCalculation = Double.NaN) {
       var classificationModel = (ISymbolicClassificationModel)model;
       var classificationProblemData = (IClassificationProblemData)problemData;
 
-      var dataset = classificationProblemData.Dataset;
-      var targetClassValues = dataset.GetDoubleValues(classificationProblemData.TargetVariable, rows);
-
-      OnlineCalculatorError errorState;
-      if (double.IsNaN(originalQuality)) {
-        var originalClassValues = classificationModel.GetEstimatedClassValues(dataset, rows);
-        originalQuality = OnlineAccuracyCalculator.Calculate(targetClassValues, originalClassValues, out errorState);
-        if (errorState != OnlineCalculatorError.None) originalQuality = 0.0;
-      }
+      if (double.IsNaN(qualityForImpactsCalculation))
+        qualityForImpactsCalculation = CalculateQualityForImpacts(classificationModel, classificationProblemData, rows);
 
       replacementValue = CalculateReplacementValue(classificationModel, node, classificationProblemData, rows);
@@ -81 +75 @@
       tempModelParentNode.InsertSubtree(i, constantNode);
 
+      OnlineCalculatorError errorState;
+      var dataset = classificationProblemData.Dataset;
+      var targetClassValues = dataset.GetDoubleValues(classificationProblemData.TargetVariable, rows);
       var estimatedClassValues = tempModel.GetEstimatedClassValues(dataset, rows);
-      double newQuality = OnlineAccuracyCalculator.Calculate(targetClassValues, estimatedClassValues, out errorState);
-      if (errorState != OnlineCalculatorError.None) newQuality = 0.0;
+      newQualityForImpactsCalculation = OnlineAccuracyCalculator.Calculate(targetClassValues, estimatedClassValues, out errorState);
+      if (errorState != OnlineCalculatorError.None) newQualityForImpactsCalculation = 0.0;
 
-      impactValue = originalQuality - newQuality;
+      impactValue = qualityForImpactsCalculation - newQualityForImpactsCalculation;
+    }
+
+    public static double CalculateQualityForImpacts(ISymbolicClassificationModel model, IClassificationProblemData problemData, IEnumerable<int> rows) {
+      OnlineCalculatorError errorState;
+      var dataset = problemData.Dataset;
+      var targetClassValues = dataset.GetDoubleValues(problemData.TargetVariable, rows);
+      var originalClassValues = model.GetEstimatedClassValues(dataset, rows);
+      var qualityForImpactsCalculation = OnlineAccuracyCalculator.Calculate(targetClassValues, originalClassValues, out errorState);
+      if (errorState != OnlineCalculatorError.None) qualityForImpactsCalculation = 0.0;
+
+      return qualityForImpactsCalculation;
     }
   }