Changeset 15950 for branches/2886_SymRegGrammarEnumeration

Timestamp:

06/05/18 14:35:03 (6 years ago)

Author:

bburlacu

Message:

#2886: Try to use variable importance information (from a random forest) to guide the search.

File:

• branches/2886_SymRegGrammarEnumeration/HeuristicLab.Algorithms.DataAnalysis.SymRegGrammarEnumeration/GrammarEnumeration/GrammarEnumerationAlgorithm.cs (modified) (9 diffs)

branches/2886_SymRegGrammarEnumeration/HeuristicLab.Algorithms.DataAnalysis.SymRegGrammarEnumeration/GrammarEnumeration/GrammarEnumerationAlgorithm.cs

@@ -4 +4 @@
 using System.Threading;
 using HeuristicLab.Algorithms.DataAnalysis.SymRegGrammarEnumeration.GrammarEnumeration;
-using HeuristicLab.Collections;
 using HeuristicLab.Common;
 using HeuristicLab.Core;
@@ -19 +18 @@
   public class GrammarEnumerationAlgorithm : FixedDataAnalysisAlgorithm<IRegressionProblem> {
     #region properties and result names
+    private readonly string VariableImportanceWeightName = "Variable Importance Weight";
     private readonly string SearchStructureSizeName = "Search Structure Size";
     private readonly string GeneratedPhrasesName = "Generated/Archived Phrases";
@@ -29 +29 @@
     private readonly string ExpansionsPerSecondName = "Expansions per second";
 
-
     private readonly string OptimizeConstantsParameterName = "Optimize Constants";
     private readonly string ErrorWeightParameterName = "Error Weight";
@@ -39 +38 @@
     public override bool SupportsPause { get { return false; } }
 
+    protected IValueParameter<DoubleValue> VariableImportanceWeightParameter {
+      get { return (IValueParameter<DoubleValue>)Parameters[VariableImportanceWeightName]; }
+    }
+
+    protected double VariableImportanceWeight {
+      get { return VariableImportanceWeightParameter.Value.Value; }
+    }
+
     protected IValueParameter<BoolValue> OptimizeConstantsParameter {
       get { return (IValueParameter<BoolValue>)Parameters[OptimizeConstantsParameterName]; }
@@ -125 +132 @@
       };
 
+      Parameters.Add(new ValueParameter<DoubleValue>(VariableImportanceWeightName, "Variable Weight.", new DoubleValue(1.0)));
       Parameters.Add(new ValueParameter<BoolValue>(OptimizeConstantsParameterName, "Run constant optimization in sentence evaluation.", new BoolValue(false)));
       Parameters.Add(new ValueParameter<DoubleValue>(ErrorWeightParameterName, "Defines, how much weight is put on a phrase's r² value when priorizing phrases during search.", new DoubleValue(0.8)));
@@ -163 +171 @@
     #endregion
 
+    private Dictionary<VariableTerminalSymbol, double> variableImportance;
+
     protected override void Run(CancellationToken cancellationToken) {
       #region init
@@ -191 +201 @@
       #endregion
 
+      #region Variable Importance
+      variableImportance = new Dictionary<VariableTerminalSymbol, double>();
+
+      RandomForestRegression rf = new RandomForestRegression();
+      rf.Problem = Problem;
+      rf.Start();
+      IRegressionSolution rfSolution = (RandomForestRegressionSolution)rf.Results["Random forest regression solution"].Value;
+      var rfImpacts = RegressionSolutionVariableImpactsCalculator.CalculateImpacts(
+        rfSolution,
+        RegressionSolutionVariableImpactsCalculator.DataPartitionEnum.Training,
+        RegressionSolutionVariableImpactsCalculator.ReplacementMethodEnum.Shuffle);
+
+      // save the normalized importances
+      var sum = rfImpacts.Sum(x => x.Item2);
+      foreach (Tuple<string, double> rfImpact in rfImpacts) {
+        VariableTerminalSymbol varSym = Grammar.VarTerminals.First(v => v.StringRepresentation == rfImpact.Item1);
+        variableImportance[varSym] = rfImpact.Item2 / sum;
+      }
+      #endregion
+
       int maxSentenceLength = GetMaxSentenceLength();
 
       OpenPhrases.Store(new SearchNode(phrase0Hash, 0.0, 0.0, phrase0));
+
+      var errorWeight = ErrorWeight;
+      var variableImportanceWeight = VariableImportanceWeight;
+
       while (OpenPhrases.Count > 0) {
         if (cancellationToken.IsCancellationRequested) break;
@@ -245 +279 @@
 
               double r2 = GetR2(newPhrase, fetchedSearchNode.R2);
-              double phrasePriority = GetPriority(newPhrase, r2, maxSentenceLength);
+              double phrasePriority = GetPriority(newPhrase, r2, maxSentenceLength, errorWeight, variableImportanceWeight);
 
               SearchNode newSearchNode = new SearchNode(phraseHash, phrasePriority, r2, newPhrase);
@@ -256 +290 @@
     }
 
-    protected double GetPriority(SymbolString phrase, double r2, int maxSentenceLength) {
+    protected double GetPriority(SymbolString phrase, double r2, int maxSentenceLength, double errorWeight, double variableImportanceWeight) {
+      var distinctVars = phrase.OfType<VariableTerminalSymbol>().Distinct();
+
+      var sum = 0d;
+      foreach (var v in distinctVars) {
+        sum += variableImportance[v];
+      }
+      var phraseVariableImportance = 1 - sum;
+
       double relLength = (double)phrase.Count() / maxSentenceLength;
       double error = 1.0 - r2;
 
-      return relLength + ErrorWeight * error;
+      return relLength + errorWeight * error + variableImportanceWeight * phraseVariableImportance;
     }