| 1 | #region License Information
|
|---|
| 2 | /* HeuristicLab
|
|---|
| 3 | * Copyright (C) 2002-2010 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
|
|---|
| 4 | *
|
|---|
| 5 | * This file is part of HeuristicLab.
|
|---|
| 6 | *
|
|---|
| 7 | * HeuristicLab is free software: you can redistribute it and/or modify
|
|---|
| 8 | * it under the terms of the GNU General Public License as published by
|
|---|
| 9 | * the Free Software Foundation, either version 3 of the License, or
|
|---|
| 10 | * (at your option) any later version.
|
|---|
| 11 | *
|
|---|
| 12 | * HeuristicLab is distributed in the hope that it will be useful,
|
|---|
| 13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|---|
| 14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|---|
| 15 | * GNU General Public License for more details.
|
|---|
| 16 | *
|
|---|
| 17 | * You should have received a copy of the GNU General Public License
|
|---|
| 18 | * along with HeuristicLab. If not, see <http://www.gnu.org/licenses/>.
|
|---|
| 19 | */
|
|---|
| 20 | #endregion
|
|---|
| 21 |
|
|---|
| 22 | using System.Collections.Generic;
|
|---|
| 23 | using System.Linq;
|
|---|
| 24 | using HeuristicLab.Core;
|
|---|
| 25 | using HeuristicLab.Data;
|
|---|
| 26 | using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
|
|---|
| 27 | using HeuristicLab.Operators;
|
|---|
| 28 | using HeuristicLab.Optimization;
|
|---|
| 29 | using HeuristicLab.Parameters;
|
|---|
| 30 | using HeuristicLab.Problems.DataAnalysis.Symbolic;
|
|---|
| 31 | using HeuristicLab.Problems.DataAnalysis.Symbolic.Symbols;
|
|---|
| 32 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
|---|
| 33 | using System;
|
|---|
| 34 |
|
|---|
| 35 | namespace HeuristicLab.Problems.DataAnalysis.Regression.Symbolic.Analyzers {
|
|---|
| 36 | public class SymbolicRegressionTournamentPruning : SingleSuccessorOperator, ISymbolicRegressionAnalyzer {
|
|---|
| 37 | private const string RandomParameterName = "Random";
|
|---|
| 38 | private const string SymbolicExpressionTreeParameterName = "SymbolicExpressionTree";
|
|---|
| 39 | private const string DataAnalysisProblemDataParameterName = "DataAnalysisProblemData";
|
|---|
| 40 | private const string SamplesStartParameterName = "SamplesStart";
|
|---|
| 41 | private const string SamplesEndParameterName = "SamplesEnd";
|
|---|
| 42 | private const string EvaluatorParameterName = "Evaluator";
|
|---|
| 43 | private const string MaximizationParameterName = "Maximization";
|
|---|
| 44 | private const string SymbolicExpressionTreeInterpreterParameterName = "SymbolicExpressionTreeInterpreter";
|
|---|
| 45 | private const string UpperEstimationLimitParameterName = "UpperEstimationLimit";
|
|---|
| 46 | private const string LowerEstimationLimitParameterName = "LowerEstimationLimit";
|
|---|
| 47 | private const string MaxPruningRatioParameterName = "MaxPruningRatio";
|
|---|
| 48 | private const string TournamentSizeParameterName = "TournamentSize";
|
|---|
| 49 | private const string PopulationPercentileStartParameterName = "PopulationPercentileStart";
|
|---|
| 50 | private const string PopulationPercentileEndParameterName = "PopulationPercentileEnd";
|
|---|
| 51 | private const string QualityGainWeightParameterName = "QualityGainWeight";
|
|---|
| 52 | private const string IterationsParameterName = "Iterations";
|
|---|
| 53 | private const string FirstPruningGenerationParameterName = "FirstPruningGeneration";
|
|---|
| 54 | private const string PruningFrequencyParameterName = "PruningFrequency";
|
|---|
| 55 | private const string GenerationParameterName = "Generations";
|
|---|
| 56 | private const string ResultsParameterName = "Results";
|
|---|
| 57 |
|
|---|
| 58 | #region parameter properties
|
|---|
| 59 | public ILookupParameter<IRandom> RandomParameter {
|
|---|
| 60 | get { return (ILookupParameter<IRandom>)Parameters[RandomParameterName]; }
|
|---|
| 61 | }
|
|---|
| 62 | public ScopeTreeLookupParameter<SymbolicExpressionTree> SymbolicExpressionTreeParameter {
|
|---|
| 63 | get { return (ScopeTreeLookupParameter<SymbolicExpressionTree>)Parameters[SymbolicExpressionTreeParameterName]; }
|
|---|
| 64 | }
|
|---|
| 65 | public ScopeTreeLookupParameter<DoubleValue> QualityParameter {
|
|---|
| 66 | get { return (ScopeTreeLookupParameter<DoubleValue>)Parameters["Quality"]; }
|
|---|
| 67 | }
|
|---|
| 68 | public ILookupParameter<DataAnalysisProblemData> DataAnalysisProblemDataParameter {
|
|---|
| 69 | get { return (ILookupParameter<DataAnalysisProblemData>)Parameters[DataAnalysisProblemDataParameterName]; }
|
|---|
| 70 | }
|
|---|
| 71 | public ILookupParameter<ISymbolicExpressionTreeInterpreter> SymbolicExpressionTreeInterpreterParameter {
|
|---|
| 72 | get { return (ILookupParameter<ISymbolicExpressionTreeInterpreter>)Parameters[SymbolicExpressionTreeInterpreterParameterName]; }
|
|---|
| 73 | }
|
|---|
| 74 | public IValueLookupParameter<DoubleValue> UpperEstimationLimitParameter {
|
|---|
| 75 | get { return (IValueLookupParameter<DoubleValue>)Parameters[UpperEstimationLimitParameterName]; }
|
|---|
| 76 | }
|
|---|
| 77 | public IValueLookupParameter<DoubleValue> LowerEstimationLimitParameter {
|
|---|
| 78 | get { return (IValueLookupParameter<DoubleValue>)Parameters[LowerEstimationLimitParameterName]; }
|
|---|
| 79 | }
|
|---|
| 80 | public IValueLookupParameter<IntValue> SamplesStartParameter {
|
|---|
| 81 | get { return (IValueLookupParameter<IntValue>)Parameters[SamplesStartParameterName]; }
|
|---|
| 82 | }
|
|---|
| 83 | public IValueLookupParameter<IntValue> SamplesEndParameter {
|
|---|
| 84 | get { return (IValueLookupParameter<IntValue>)Parameters[SamplesEndParameterName]; }
|
|---|
| 85 | }
|
|---|
| 86 | public IValueLookupParameter<PercentValue> RelativeNumberOfEvaluatedRowsParameters {
|
|---|
| 87 | get { return (IValueLookupParameter<PercentValue>)Parameters["RelativeNumberOfEvaluatedRows"]; }
|
|---|
| 88 | }
|
|---|
| 89 | public ILookupParameter<ISymbolicRegressionEvaluator> EvaluatorParameter {
|
|---|
| 90 | get { return (ILookupParameter<ISymbolicRegressionEvaluator>)Parameters[EvaluatorParameterName]; }
|
|---|
| 91 | }
|
|---|
| 92 | public ILookupParameter<BoolValue> MaximizationParameter {
|
|---|
| 93 | get { return (ILookupParameter<BoolValue>)Parameters[MaximizationParameterName]; }
|
|---|
| 94 | }
|
|---|
| 95 | public IValueLookupParameter<DoubleValue> MaxPruningRatioParameter {
|
|---|
| 96 | get { return (IValueLookupParameter<DoubleValue>)Parameters[MaxPruningRatioParameterName]; }
|
|---|
| 97 | }
|
|---|
| 98 | public IValueLookupParameter<IntValue> TournamentSizeParameter {
|
|---|
| 99 | get { return (IValueLookupParameter<IntValue>)Parameters[TournamentSizeParameterName]; }
|
|---|
| 100 | }
|
|---|
| 101 | public IValueLookupParameter<DoubleValue> PopulationPercentileStartParameter {
|
|---|
| 102 | get { return (IValueLookupParameter<DoubleValue>)Parameters[PopulationPercentileStartParameterName]; }
|
|---|
| 103 | }
|
|---|
| 104 | public IValueLookupParameter<DoubleValue> PopulationPercentileEndParameter {
|
|---|
| 105 | get { return (IValueLookupParameter<DoubleValue>)Parameters[PopulationPercentileEndParameterName]; }
|
|---|
| 106 | }
|
|---|
| 107 | public IValueLookupParameter<DoubleValue> QualityGainWeightParameter {
|
|---|
| 108 | get { return (IValueLookupParameter<DoubleValue>)Parameters[QualityGainWeightParameterName]; }
|
|---|
| 109 | }
|
|---|
| 110 | public IValueLookupParameter<IntValue> IterationsParameter {
|
|---|
| 111 | get { return (IValueLookupParameter<IntValue>)Parameters[IterationsParameterName]; }
|
|---|
| 112 | }
|
|---|
| 113 | public IValueLookupParameter<IntValue> FirstPruningGenerationParameter {
|
|---|
| 114 | get { return (IValueLookupParameter<IntValue>)Parameters[FirstPruningGenerationParameterName]; }
|
|---|
| 115 | }
|
|---|
| 116 | public IValueLookupParameter<IntValue> PruningFrequencyParameter {
|
|---|
| 117 | get { return (IValueLookupParameter<IntValue>)Parameters[PruningFrequencyParameterName]; }
|
|---|
| 118 | }
|
|---|
| 119 | public ILookupParameter<IntValue> GenerationParameter {
|
|---|
| 120 | get { return (ILookupParameter<IntValue>)Parameters[GenerationParameterName]; }
|
|---|
| 121 | }
|
|---|
| 122 | public ILookupParameter<ResultCollection> ResultsParameter {
|
|---|
| 123 | get { return (ILookupParameter<ResultCollection>)Parameters[ResultsParameterName]; }
|
|---|
| 124 | }
|
|---|
| 125 | public IValueLookupParameter<BoolValue> ApplyPruningParameter {
|
|---|
| 126 | get { return (IValueLookupParameter<BoolValue>)Parameters["ApplyPruning"]; }
|
|---|
| 127 | }
|
|---|
| 128 | #endregion
|
|---|
| 129 | #region properties
|
|---|
| 130 | public IRandom Random {
|
|---|
| 131 | get { return RandomParameter.ActualValue; }
|
|---|
| 132 | }
|
|---|
| 133 | public ItemArray<SymbolicExpressionTree> SymbolicExpressionTree {
|
|---|
| 134 | get { return SymbolicExpressionTreeParameter.ActualValue; }
|
|---|
| 135 | }
|
|---|
| 136 | public DataAnalysisProblemData DataAnalysisProblemData {
|
|---|
| 137 | get { return DataAnalysisProblemDataParameter.ActualValue; }
|
|---|
| 138 | }
|
|---|
| 139 | public ISymbolicExpressionTreeInterpreter SymbolicExpressionTreeInterpreter {
|
|---|
| 140 | get { return SymbolicExpressionTreeInterpreterParameter.ActualValue; }
|
|---|
| 141 | }
|
|---|
| 142 | public DoubleValue UpperEstimationLimit {
|
|---|
| 143 | get { return UpperEstimationLimitParameter.ActualValue; }
|
|---|
| 144 | }
|
|---|
| 145 | public DoubleValue LowerEstimationLimit {
|
|---|
| 146 | get { return LowerEstimationLimitParameter.ActualValue; }
|
|---|
| 147 | }
|
|---|
| 148 | public IntValue SamplesStart {
|
|---|
| 149 | get { return SamplesStartParameter.ActualValue; }
|
|---|
| 150 | }
|
|---|
| 151 | public IntValue SamplesEnd {
|
|---|
| 152 | get { return SamplesEndParameter.ActualValue; }
|
|---|
| 153 | }
|
|---|
| 154 | public ISymbolicRegressionEvaluator Evaluator {
|
|---|
| 155 | get { return EvaluatorParameter.ActualValue; }
|
|---|
| 156 | }
|
|---|
| 157 | public BoolValue Maximization {
|
|---|
| 158 | get { return MaximizationParameter.ActualValue; }
|
|---|
| 159 | }
|
|---|
| 160 | public DoubleValue MaxPruningRatio {
|
|---|
| 161 | get { return MaxPruningRatioParameter.ActualValue; }
|
|---|
| 162 | }
|
|---|
| 163 | public IntValue TournamentSize {
|
|---|
| 164 | get { return TournamentSizeParameter.ActualValue; }
|
|---|
| 165 | }
|
|---|
| 166 | public DoubleValue PopulationPercentileStart {
|
|---|
| 167 | get { return PopulationPercentileStartParameter.ActualValue; }
|
|---|
| 168 | }
|
|---|
| 169 | public DoubleValue PopulationPercentileEnd {
|
|---|
| 170 | get { return PopulationPercentileEndParameter.ActualValue; }
|
|---|
| 171 | }
|
|---|
| 172 | public DoubleValue QualityGainWeight {
|
|---|
| 173 | get { return QualityGainWeightParameter.ActualValue; }
|
|---|
| 174 | }
|
|---|
| 175 | public IntValue Iterations {
|
|---|
| 176 | get { return IterationsParameter.ActualValue; }
|
|---|
| 177 | }
|
|---|
| 178 | public IntValue PruningFrequency {
|
|---|
| 179 | get { return PruningFrequencyParameter.ActualValue; }
|
|---|
| 180 | }
|
|---|
| 181 | public IntValue FirstPruningGeneration {
|
|---|
| 182 | get { return FirstPruningGenerationParameter.ActualValue; }
|
|---|
| 183 | }
|
|---|
| 184 | public IntValue Generation {
|
|---|
| 185 | get { return GenerationParameter.ActualValue; }
|
|---|
| 186 | }
|
|---|
| 187 | #endregion
|
|---|
| 188 | [StorableConstructor]
|
|---|
| 189 | protected SymbolicRegressionTournamentPruning(bool deserializing) : base(deserializing) { }
|
|---|
| 190 | public SymbolicRegressionTournamentPruning()
|
|---|
| 191 | : base() {
|
|---|
| 192 | Parameters.Add(new LookupParameter<IRandom>(RandomParameterName, "A random number generator."));
|
|---|
| 193 | Parameters.Add(new ScopeTreeLookupParameter<SymbolicExpressionTree>(SymbolicExpressionTreeParameterName, "The symbolic expression trees to prune."));
|
|---|
| 194 | Parameters.Add(new ScopeTreeLookupParameter<DoubleValue>("Quality"));
|
|---|
| 195 | Parameters.Add(new LookupParameter<DataAnalysisProblemData>(DataAnalysisProblemDataParameterName, "The data analysis problem data to use for branch impact evaluation."));
|
|---|
| 196 | Parameters.Add(new LookupParameter<ISymbolicExpressionTreeInterpreter>(SymbolicExpressionTreeInterpreterParameterName, "The interpreter to use for node impact evaluation"));
|
|---|
| 197 | Parameters.Add(new ValueLookupParameter<IntValue>(SamplesStartParameterName, "The first row index of the dataset partition to use for branch impact evaluation."));
|
|---|
| 198 | Parameters.Add(new ValueLookupParameter<IntValue>(SamplesEndParameterName, "The last row index of the dataset partition to use for branch impact evaluation."));
|
|---|
| 199 | Parameters.Add(new LookupParameter<ISymbolicRegressionEvaluator>(EvaluatorParameterName, "The evaluator that should be used to determine which branches are not relevant."));
|
|---|
| 200 | Parameters.Add(new LookupParameter<BoolValue>(MaximizationParameterName, "The direction of optimization."));
|
|---|
| 201 | Parameters.Add(new ValueLookupParameter<BoolValue>("ApplyPruning"));
|
|---|
| 202 | Parameters.Add(new ValueLookupParameter<DoubleValue>(MaxPruningRatioParameterName, "The maximal relative size of the pruned branch.", new DoubleValue(0.5)));
|
|---|
| 203 | Parameters.Add(new ValueLookupParameter<IntValue>(TournamentSizeParameterName, "The number of branches to compare for pruning", new IntValue(10)));
|
|---|
| 204 | Parameters.Add(new ValueLookupParameter<DoubleValue>(PopulationPercentileStartParameterName, "The start of the population percentile to consider for pruning.", new DoubleValue(0.25)));
|
|---|
| 205 | Parameters.Add(new ValueLookupParameter<DoubleValue>(PopulationPercentileEndParameterName, "The end of the population percentile to consider for pruning.", new DoubleValue(0.75)));
|
|---|
| 206 | Parameters.Add(new ValueLookupParameter<DoubleValue>(QualityGainWeightParameterName, "The weight of the quality gain relative to the size gain.", new DoubleValue(1.0)));
|
|---|
| 207 | Parameters.Add(new ValueLookupParameter<DoubleValue>(UpperEstimationLimitParameterName, "The upper estimation limit to use for evaluation."));
|
|---|
| 208 | Parameters.Add(new ValueLookupParameter<DoubleValue>(LowerEstimationLimitParameterName, "The lower estimation limit to use for evaluation."));
|
|---|
| 209 | Parameters.Add(new ValueLookupParameter<IntValue>(IterationsParameterName, "The number of pruning iterations to apply for each tree.", new IntValue(1)));
|
|---|
| 210 | Parameters.Add(new ValueLookupParameter<IntValue>(FirstPruningGenerationParameterName, "The first generation when pruning should be applied.", new IntValue(1)));
|
|---|
| 211 | Parameters.Add(new ValueLookupParameter<IntValue>(PruningFrequencyParameterName, "The frequency of pruning operations (1: every generation, 2: every second generation...)", new IntValue(1)));
|
|---|
| 212 | Parameters.Add(new LookupParameter<IntValue>(GenerationParameterName, "The current generation."));
|
|---|
| 213 | Parameters.Add(new LookupParameter<ResultCollection>(ResultsParameterName, "The results collection."));
|
|---|
| 214 | Parameters.Add(new ValueLookupParameter<PercentValue>("RelativeNumberOfEvaluatedRows", new PercentValue(1.0)));
|
|---|
| 215 | }
|
|---|
| 216 |
|
|---|
| 217 | [StorableHook(HookType.AfterDeserialization)]
|
|---|
| 218 | private void AfterDeserialization() {
|
|---|
| 219 | #region compatibility remove before releasing 3.3.1
|
|---|
| 220 | if (!Parameters.ContainsKey(EvaluatorParameterName)) {
|
|---|
| 221 | Parameters.Add(new LookupParameter<ISymbolicRegressionEvaluator>(EvaluatorParameterName, "The evaluator which should be used to evaluate the solution on the validation set."));
|
|---|
| 222 | }
|
|---|
| 223 | if (!Parameters.ContainsKey(MaximizationParameterName)) {
|
|---|
| 224 | Parameters.Add(new LookupParameter<BoolValue>(MaximizationParameterName, "The direction of optimization."));
|
|---|
| 225 | }
|
|---|
| 226 | if (!Parameters.ContainsKey("ApplyPruning")) {
|
|---|
| 227 | Parameters.Add(new ValueLookupParameter<BoolValue>("ApplyPruning"));
|
|---|
| 228 | }
|
|---|
| 229 | if (!Parameters.ContainsKey("Quality")) {
|
|---|
| 230 | Parameters.Add(new ScopeTreeLookupParameter<DoubleValue>("Quality"));
|
|---|
| 231 | }
|
|---|
| 232 | if (!Parameters.ContainsKey("RelativeNumberOfEvaluatedRows")) {
|
|---|
| 233 | Parameters.Add(new ValueLookupParameter<PercentValue>("RelativeNumberOfEvaluatedRows", new PercentValue(1.0)));
|
|---|
| 234 | }
|
|---|
| 235 |
|
|---|
| 236 | #endregion
|
|---|
| 237 | }
|
|---|
| 238 |
|
|---|
| 239 | public override IOperation Apply() {
|
|---|
| 240 | bool pruningCondition =
|
|---|
| 241 | (ApplyPruningParameter.ActualValue.Value) &&
|
|---|
| 242 | (Generation.Value >= FirstPruningGeneration.Value) &&
|
|---|
| 243 | ((Generation.Value - FirstPruningGeneration.Value) % PruningFrequency.Value == 0);
|
|---|
| 244 | if (pruningCondition) {
|
|---|
| 245 | int n = SymbolicExpressionTree.Length;
|
|---|
| 246 | double percentileStart = PopulationPercentileStart.Value;
|
|---|
| 247 | double percentileEnd = PopulationPercentileEnd.Value;
|
|---|
| 248 | // for each tree in the given percentile
|
|---|
| 249 | ItemArray<SymbolicExpressionTree> trees = SymbolicExpressionTree;
|
|---|
| 250 | ItemArray<DoubleValue> quality = QualityParameter.ActualValue;
|
|---|
| 251 | bool maximization = Maximization.Value;
|
|---|
| 252 | var selectedTrees = (from index in Enumerable.Range(0, n)
|
|---|
| 253 | orderby maximization ? -quality[index].Value : quality[index].Value
|
|---|
| 254 | select new { Tree = trees[index], Quality = quality[index] })
|
|---|
| 255 | .Skip((int)(n * percentileStart))
|
|---|
| 256 | .Take((int)(n * (percentileEnd - percentileStart)));
|
|---|
| 257 | foreach (var pair in selectedTrees) {
|
|---|
| 258 | Prune(Random, pair.Tree, pair.Quality, Iterations.Value, TournamentSize.Value,
|
|---|
| 259 | DataAnalysisProblemData, SamplesStart.Value, SamplesEnd.Value, RelativeNumberOfEvaluatedRowsParameters.ActualValue.Value,
|
|---|
| 260 | SymbolicExpressionTreeInterpreter, Evaluator, Maximization.Value,
|
|---|
| 261 | LowerEstimationLimit.Value, UpperEstimationLimit.Value,
|
|---|
| 262 | MaxPruningRatio.Value, QualityGainWeight.Value);
|
|---|
| 263 | }
|
|---|
| 264 | }
|
|---|
| 265 | return base.Apply();
|
|---|
| 266 | }
|
|---|
| 267 |
|
|---|
| 268 | public static void Prune(IRandom random, SymbolicExpressionTree tree, DoubleValue quality, int iterations, int tournamentSize,
|
|---|
| 269 | DataAnalysisProblemData problemData, int samplesStart, int samplesEnd, double relativeNumberOfEvaluatedRows,
|
|---|
| 270 | ISymbolicExpressionTreeInterpreter interpreter, ISymbolicRegressionEvaluator evaluator, bool maximization,
|
|---|
| 271 | double lowerEstimationLimit, double upperEstimationLimit,
|
|---|
| 272 | double maxPruningRatio, double qualityGainWeight) {
|
|---|
| 273 |
|
|---|
| 274 | IEnumerable<int> rows = RandomEnumerable.SampleRandomNumbers(samplesStart, samplesEnd, (int)Math.Ceiling((samplesEnd - samplesStart) * relativeNumberOfEvaluatedRows));
|
|---|
| 275 | int originalSize = tree.Size;
|
|---|
| 276 |
|
|---|
| 277 | int minPrunedSize = (int)(originalSize * (1 - maxPruningRatio));
|
|---|
| 278 | // tree for branch evaluation
|
|---|
| 279 | SymbolicExpressionTree templateTree = (SymbolicExpressionTree)tree.Clone();
|
|---|
| 280 | while (templateTree.Root.SubTrees[0].SubTrees.Count > 0) templateTree.Root.SubTrees[0].RemoveSubTree(0);
|
|---|
| 281 |
|
|---|
| 282 | SymbolicExpressionTree prunedTree = tree;
|
|---|
| 283 | double currentQuality = quality.Value;
|
|---|
| 284 | for (int iteration = 0; iteration < iterations; iteration++) {
|
|---|
| 285 | SymbolicExpressionTree iterationBestTree = prunedTree;
|
|---|
| 286 | double bestGain = double.PositiveInfinity;
|
|---|
| 287 | int maxPrunedBranchSize = (int)(prunedTree.Size * maxPruningRatio);
|
|---|
| 288 |
|
|---|
| 289 | for (int i = 0; i < tournamentSize; i++) {
|
|---|
| 290 | var clonedTree = (SymbolicExpressionTree)prunedTree.Clone();
|
|---|
| 291 | int clonedTreeSize = clonedTree.Size;
|
|---|
| 292 | var prunePoints = (from node in clonedTree.Root.SubTrees[0].IterateNodesPostfix()
|
|---|
| 293 | from subTree in node.SubTrees
|
|---|
| 294 | let subTreeSize = subTree.GetSize()
|
|---|
| 295 | where subTreeSize <= maxPrunedBranchSize
|
|---|
| 296 | where clonedTreeSize - subTreeSize >= minPrunedSize
|
|---|
| 297 | select new { Parent = node, Branch = subTree, SubTreeIndex = node.SubTrees.IndexOf(subTree) })
|
|---|
| 298 | .ToList();
|
|---|
| 299 | if (prunePoints.Count > 0) {
|
|---|
| 300 | var selectedPrunePoint = prunePoints.SelectRandom(random);
|
|---|
| 301 | templateTree.Root.SubTrees[0].AddSubTree(selectedPrunePoint.Branch);
|
|---|
| 302 | IEnumerable<double> branchValues = interpreter.GetSymbolicExpressionTreeValues(templateTree, problemData.Dataset, rows);
|
|---|
| 303 | double branchMean = branchValues.Average();
|
|---|
| 304 | templateTree.Root.SubTrees[0].RemoveSubTree(0);
|
|---|
| 305 |
|
|---|
| 306 | selectedPrunePoint.Parent.RemoveSubTree(selectedPrunePoint.SubTreeIndex);
|
|---|
| 307 | var constNode = CreateConstant(branchMean);
|
|---|
| 308 | selectedPrunePoint.Parent.InsertSubTree(selectedPrunePoint.SubTreeIndex, constNode);
|
|---|
| 309 |
|
|---|
| 310 | double prunedQuality = evaluator.Evaluate(interpreter, clonedTree,
|
|---|
| 311 | lowerEstimationLimit, upperEstimationLimit, problemData.Dataset, problemData.TargetVariable.Value, rows);
|
|---|
| 312 | double prunedSize = clonedTree.Size;
|
|---|
| 313 | // deteriation in quality:
|
|---|
| 314 | // exp: MSE : newMse < origMse (improvement) => prefer the larger improvement
|
|---|
| 315 | // MSE : newMse > origMse (deteriation) => prefer the smaller deteriation
|
|---|
| 316 | // MSE : minimize: newMse / origMse
|
|---|
| 317 | // R² : newR² > origR² (improvment) => prefer the larger improvment
|
|---|
| 318 | // R² : newR² < origR² (deteriation) => prefer smaller deteriation
|
|---|
| 319 | // R² : minimize: origR² / newR²
|
|---|
| 320 | double qualityDeteriation = maximization ? quality.Value / prunedQuality : prunedQuality / quality.Value;
|
|---|
| 321 | // size of the pruned tree is always smaller than the size of the original tree
|
|---|
| 322 | // same change in quality => prefer pruning operation that removes a larger tree
|
|---|
| 323 | double gain = (qualityDeteriation * qualityGainWeight) /
|
|---|
| 324 | (originalSize / prunedSize);
|
|---|
| 325 | if (gain < bestGain) {
|
|---|
| 326 | bestGain = gain;
|
|---|
| 327 | iterationBestTree = clonedTree;
|
|---|
| 328 | currentQuality = prunedQuality;
|
|---|
| 329 | }
|
|---|
| 330 | }
|
|---|
| 331 | }
|
|---|
| 332 | prunedTree = iterationBestTree;
|
|---|
| 333 | }
|
|---|
| 334 |
|
|---|
| 335 | quality.Value = currentQuality;
|
|---|
| 336 | tree.Root = prunedTree.Root;
|
|---|
| 337 | }
|
|---|
| 338 |
|
|---|
| 339 | private static SymbolicExpressionTreeNode CreateConstant(double constantValue) {
|
|---|
| 340 | var node = (ConstantTreeNode)(new Constant()).CreateTreeNode();
|
|---|
| 341 | node.Value = constantValue;
|
|---|
| 342 | return node;
|
|---|
| 343 | }
|
|---|
| 344 | }
|
|---|
| 345 | }
|
|---|
