| 1 | #region License Information
|
|---|
| 2 | /* HeuristicLab
|
|---|
| 3 | * Copyright (C) 2002-2011 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.Analysis;
|
|---|
| 25 | using HeuristicLab.Common;
|
|---|
| 26 | using HeuristicLab.Core;
|
|---|
| 27 | using HeuristicLab.Data;
|
|---|
| 28 | using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
|
|---|
| 29 | using HeuristicLab.Operators;
|
|---|
| 30 | using HeuristicLab.Optimization;
|
|---|
| 31 | using HeuristicLab.Parameters;
|
|---|
| 32 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
|---|
| 33 | using HeuristicLab.PluginInfrastructure;
|
|---|
| 34 | using HeuristicLab.Problems.DataAnalysis.Regression.Symbolic;
|
|---|
| 35 | using HeuristicLab.Problems.DataAnalysis.Regression.Symbolic.Analyzers;
|
|---|
| 36 | using HeuristicLab.Problems.DataAnalysis.Symbolic;
|
|---|
| 37 |
|
|---|
| 38 | namespace HeuristicLab.Problems.DataAnalysis.Classification {
|
|---|
| 39 | [Item("ValidationBestSymbolicClassificationSolutionAnalyzer", "An operator that analyzes the validation best symbolic classification solution.")]
|
|---|
| 40 | [StorableClass]
|
|---|
| 41 | [NonDiscoverableType]
|
|---|
| 42 | public class ValidationBestSymbolicClassificationSolutionAnalyzer : SingleSuccessorOperator, ISymbolicClassificationAnalyzer {
|
|---|
| 43 | private const string MaximizationParameterName = "Maximization";
|
|---|
| 44 | private const string GenerationsParameterName = "Generations";
|
|---|
| 45 | private const string RandomParameterName = "Random";
|
|---|
| 46 | private const string SymbolicExpressionTreeParameterName = "SymbolicExpressionTree";
|
|---|
| 47 | private const string SymbolicExpressionTreeInterpreterParameterName = "SymbolicExpressionTreeInterpreter";
|
|---|
| 48 |
|
|---|
| 49 | private const string ClassificationProblemDataParameterName = "ClassificationProblemData";
|
|---|
| 50 | private const string EvaluatorParameterName = "Evaluator";
|
|---|
| 51 | private const string ValidationSamplesStartParameterName = "SamplesStart";
|
|---|
| 52 | private const string ValidationSamplesEndParameterName = "SamplesEnd";
|
|---|
| 53 | private const string RelativeNumberOfEvaluatedSamplesParameterName = "RelativeNumberOfEvaluatedSamples";
|
|---|
| 54 | private const string UpperEstimationLimitParameterName = "UpperEstimationLimit";
|
|---|
| 55 | private const string LowerEstimationLimitParameterName = "LowerEstimationLimit";
|
|---|
| 56 | private const string CalculateSolutionComplexityParameterName = "CalculateSolutionComplexity";
|
|---|
| 57 | private const string ApplyLinearScalingParameterName = "ApplyLinearScaling";
|
|---|
| 58 |
|
|---|
| 59 | private const string ResultsParameterName = "Results";
|
|---|
| 60 | private const string BestValidationQualityParameterName = "Best validation quality";
|
|---|
| 61 | private const string BestValidationSolutionParameterName = "Best validation solution";
|
|---|
| 62 | private const string BestSolutionAccuracyTrainingParameterName = "Best solution accuracy (training)";
|
|---|
| 63 | private const string BestSolutionAccuracyTestParameterName = "Best solution accuracy (test)";
|
|---|
| 64 | private const string BestSolutionLengthParameterName = "Best solution length (on validation set)";
|
|---|
| 65 | private const string BestSolutionHeightParameterName = "Best solution height (on validation set)";
|
|---|
| 66 | private const string VariableFrequenciesParameterName = "VariableFrequencies";
|
|---|
| 67 |
|
|---|
| 68 | #region parameter properties
|
|---|
| 69 | public ILookupParameter<BoolValue> MaximizationParameter {
|
|---|
| 70 | get { return (ILookupParameter<BoolValue>)Parameters[MaximizationParameterName]; }
|
|---|
| 71 | }
|
|---|
| 72 | public ILookupParameter<IntValue> GenerationsParameter {
|
|---|
| 73 | get { return (ILookupParameter<IntValue>)Parameters[GenerationsParameterName]; }
|
|---|
| 74 | }
|
|---|
| 75 | public ILookupParameter<IRandom> RandomParameter {
|
|---|
| 76 | get { return (ILookupParameter<IRandom>)Parameters[RandomParameterName]; }
|
|---|
| 77 | }
|
|---|
| 78 | public ScopeTreeLookupParameter<SymbolicExpressionTree> SymbolicExpressionTreeParameter {
|
|---|
| 79 | get { return (ScopeTreeLookupParameter<SymbolicExpressionTree>)Parameters[SymbolicExpressionTreeParameterName]; }
|
|---|
| 80 | }
|
|---|
| 81 | public IValueLookupParameter<ISymbolicExpressionTreeInterpreter> SymbolicExpressionTreeInterpreterParameter {
|
|---|
| 82 | get { return (IValueLookupParameter<ISymbolicExpressionTreeInterpreter>)Parameters[SymbolicExpressionTreeInterpreterParameterName]; }
|
|---|
| 83 | }
|
|---|
| 84 | public ILookupParameter<ClassificationProblemData> ClassificationProblemDataParameter {
|
|---|
| 85 | get { return (ILookupParameter<ClassificationProblemData>)Parameters[ClassificationProblemDataParameterName]; }
|
|---|
| 86 | }
|
|---|
| 87 | public ILookupParameter<ISymbolicClassificationEvaluator> EvaluatorParameter {
|
|---|
| 88 | get { return (ILookupParameter<ISymbolicClassificationEvaluator>)Parameters[EvaluatorParameterName]; }
|
|---|
| 89 | }
|
|---|
| 90 | public IValueLookupParameter<IntValue> ValidationSamplesStartParameter {
|
|---|
| 91 | get { return (IValueLookupParameter<IntValue>)Parameters[ValidationSamplesStartParameterName]; }
|
|---|
| 92 | }
|
|---|
| 93 | public IValueLookupParameter<IntValue> ValidationSamplesEndParameter {
|
|---|
| 94 | get { return (IValueLookupParameter<IntValue>)Parameters[ValidationSamplesEndParameterName]; }
|
|---|
| 95 | }
|
|---|
| 96 | public IValueParameter<PercentValue> RelativeNumberOfEvaluatedSamplesParameter {
|
|---|
| 97 | get { return (IValueParameter<PercentValue>)Parameters[RelativeNumberOfEvaluatedSamplesParameterName]; }
|
|---|
| 98 | }
|
|---|
| 99 | public IValueLookupParameter<DoubleValue> UpperEstimationLimitParameter {
|
|---|
| 100 | get { return (IValueLookupParameter<DoubleValue>)Parameters[UpperEstimationLimitParameterName]; }
|
|---|
| 101 | }
|
|---|
| 102 | public IValueLookupParameter<DoubleValue> LowerEstimationLimitParameter {
|
|---|
| 103 | get { return (IValueLookupParameter<DoubleValue>)Parameters[LowerEstimationLimitParameterName]; }
|
|---|
| 104 | }
|
|---|
| 105 | public IValueLookupParameter<BoolValue> ApplyLinearScalingParameter {
|
|---|
| 106 | get { return (IValueLookupParameter<BoolValue>)Parameters[ApplyLinearScalingParameterName]; }
|
|---|
| 107 | }
|
|---|
| 108 | public ILookupParameter<DataTable> VariableFrequenciesParameter {
|
|---|
| 109 | get { return (ILookupParameter<DataTable>)Parameters[VariableFrequenciesParameterName]; }
|
|---|
| 110 | }
|
|---|
| 111 | public IValueParameter<BoolValue> CalculateSolutionComplexityParameter {
|
|---|
| 112 | get { return (IValueParameter<BoolValue>)Parameters[CalculateSolutionComplexityParameterName]; }
|
|---|
| 113 | }
|
|---|
| 114 |
|
|---|
| 115 | public ILookupParameter<ResultCollection> ResultsParameter {
|
|---|
| 116 | get { return (ILookupParameter<ResultCollection>)Parameters[ResultsParameterName]; }
|
|---|
| 117 | }
|
|---|
| 118 | public ILookupParameter<DoubleValue> BestValidationQualityParameter {
|
|---|
| 119 | get { return (ILookupParameter<DoubleValue>)Parameters[BestValidationQualityParameterName]; }
|
|---|
| 120 | }
|
|---|
| 121 | public ILookupParameter<SymbolicClassificationSolution> BestValidationSolutionParameter {
|
|---|
| 122 | get { return (ILookupParameter<SymbolicClassificationSolution>)Parameters[BestValidationSolutionParameterName]; }
|
|---|
| 123 | }
|
|---|
| 124 | public ILookupParameter<DoubleValue> BestSolutionAccuracyTrainingParameter {
|
|---|
| 125 | get { return (ILookupParameter<DoubleValue>)Parameters[BestSolutionAccuracyTrainingParameterName]; }
|
|---|
| 126 | }
|
|---|
| 127 | public ILookupParameter<DoubleValue> BestSolutionAccuracyTestParameter {
|
|---|
| 128 | get { return (ILookupParameter<DoubleValue>)Parameters[BestSolutionAccuracyTestParameterName]; }
|
|---|
| 129 | }
|
|---|
| 130 | public ILookupParameter<IntValue> BestSolutionLengthParameter {
|
|---|
| 131 | get { return (ILookupParameter<IntValue>)Parameters[BestSolutionLengthParameterName]; }
|
|---|
| 132 | }
|
|---|
| 133 | public ILookupParameter<IntValue> BestSolutionHeightParameter {
|
|---|
| 134 | get { return (ILookupParameter<IntValue>)Parameters[BestSolutionHeightParameterName]; }
|
|---|
| 135 | }
|
|---|
| 136 | #endregion
|
|---|
| 137 | #region properties
|
|---|
| 138 | public BoolValue Maximization {
|
|---|
| 139 | get { return MaximizationParameter.ActualValue; }
|
|---|
| 140 | }
|
|---|
| 141 | public IntValue Generations {
|
|---|
| 142 | get { return GenerationsParameter.ActualValue; }
|
|---|
| 143 | }
|
|---|
| 144 | public IRandom Random {
|
|---|
| 145 | get { return RandomParameter.ActualValue; }
|
|---|
| 146 | }
|
|---|
| 147 | public ItemArray<SymbolicExpressionTree> SymbolicExpressionTree {
|
|---|
| 148 | get { return SymbolicExpressionTreeParameter.ActualValue; }
|
|---|
| 149 | }
|
|---|
| 150 | public ISymbolicExpressionTreeInterpreter SymbolicExpressionTreeInterpreter {
|
|---|
| 151 | get { return SymbolicExpressionTreeInterpreterParameter.ActualValue; }
|
|---|
| 152 | }
|
|---|
| 153 |
|
|---|
| 154 | public ClassificationProblemData ClassificationProblemData {
|
|---|
| 155 | get { return ClassificationProblemDataParameter.ActualValue; }
|
|---|
| 156 | }
|
|---|
| 157 | public ISymbolicClassificationEvaluator Evaluator {
|
|---|
| 158 | get { return EvaluatorParameter.ActualValue; }
|
|---|
| 159 | }
|
|---|
| 160 | public IntValue ValidiationSamplesStart {
|
|---|
| 161 | get { return ValidationSamplesStartParameter.ActualValue; }
|
|---|
| 162 | }
|
|---|
| 163 | public IntValue ValidationSamplesEnd {
|
|---|
| 164 | get { return ValidationSamplesEndParameter.ActualValue; }
|
|---|
| 165 | }
|
|---|
| 166 | public PercentValue RelativeNumberOfEvaluatedSamples {
|
|---|
| 167 | get { return RelativeNumberOfEvaluatedSamplesParameter.Value; }
|
|---|
| 168 | }
|
|---|
| 169 | public DoubleValue UpperEstimationLimit {
|
|---|
| 170 | get { return UpperEstimationLimitParameter.ActualValue; }
|
|---|
| 171 | }
|
|---|
| 172 | public DoubleValue LowerEstimationLimit {
|
|---|
| 173 | get { return LowerEstimationLimitParameter.ActualValue; }
|
|---|
| 174 | }
|
|---|
| 175 | public BoolValue ApplyLinearScaling {
|
|---|
| 176 | get { return ApplyLinearScalingParameter.ActualValue; }
|
|---|
| 177 | set { ApplyLinearScalingParameter.ActualValue = value; }
|
|---|
| 178 | }
|
|---|
| 179 | public DataTable VariableFrequencies {
|
|---|
| 180 | get { return VariableFrequenciesParameter.ActualValue; }
|
|---|
| 181 | }
|
|---|
| 182 | public BoolValue CalculateSolutionComplexity {
|
|---|
| 183 | get { return CalculateSolutionComplexityParameter.Value; }
|
|---|
| 184 | set { CalculateSolutionComplexityParameter.Value = value; }
|
|---|
| 185 | }
|
|---|
| 186 |
|
|---|
| 187 | public ResultCollection Results {
|
|---|
| 188 | get { return ResultsParameter.ActualValue; }
|
|---|
| 189 | }
|
|---|
| 190 | public DoubleValue BestValidationQuality {
|
|---|
| 191 | get { return BestValidationQualityParameter.ActualValue; }
|
|---|
| 192 | protected set { BestValidationQualityParameter.ActualValue = value; }
|
|---|
| 193 | }
|
|---|
| 194 | public SymbolicClassificationSolution BestValidationSolution {
|
|---|
| 195 | get { return BestValidationSolutionParameter.ActualValue; }
|
|---|
| 196 | protected set { BestValidationSolutionParameter.ActualValue = value; }
|
|---|
| 197 | }
|
|---|
| 198 | public DoubleValue BestSolutionAccuracyTraining {
|
|---|
| 199 | get { return BestSolutionAccuracyTrainingParameter.ActualValue; }
|
|---|
| 200 | protected set { BestSolutionAccuracyTrainingParameter.ActualValue = value; }
|
|---|
| 201 | }
|
|---|
| 202 | public DoubleValue BestSolutionAccuracyTest {
|
|---|
| 203 | get { return BestSolutionAccuracyTestParameter.ActualValue; }
|
|---|
| 204 | protected set { BestSolutionAccuracyTestParameter.ActualValue = value; }
|
|---|
| 205 | }
|
|---|
| 206 | public IntValue BestSolutionLength {
|
|---|
| 207 | get { return BestSolutionLengthParameter.ActualValue; }
|
|---|
| 208 | set { BestSolutionLengthParameter.ActualValue = value; }
|
|---|
| 209 | }
|
|---|
| 210 | public IntValue BestSolutionHeight {
|
|---|
| 211 | get { return BestSolutionHeightParameter.ActualValue; }
|
|---|
| 212 | set { BestSolutionHeightParameter.ActualValue = value; }
|
|---|
| 213 | }
|
|---|
| 214 | #endregion
|
|---|
| 215 |
|
|---|
| 216 | [StorableConstructor]
|
|---|
| 217 | protected ValidationBestSymbolicClassificationSolutionAnalyzer(bool deserializing) : base(deserializing) { }
|
|---|
| 218 | protected ValidationBestSymbolicClassificationSolutionAnalyzer(ValidationBestSymbolicClassificationSolutionAnalyzer original, Cloner cloner)
|
|---|
| 219 | : base(original, cloner) {
|
|---|
| 220 | }
|
|---|
| 221 | public ValidationBestSymbolicClassificationSolutionAnalyzer()
|
|---|
| 222 | : base() {
|
|---|
| 223 | Parameters.Add(new LookupParameter<BoolValue>(MaximizationParameterName, "The direction of optimization."));
|
|---|
| 224 | Parameters.Add(new LookupParameter<IntValue>(GenerationsParameterName, "The number of generations calculated so far."));
|
|---|
| 225 | Parameters.Add(new LookupParameter<IRandom>(RandomParameterName, "The random generator to use."));
|
|---|
| 226 | Parameters.Add(new ScopeTreeLookupParameter<SymbolicExpressionTree>(SymbolicExpressionTreeParameterName, "The symbolic expression trees to analyze."));
|
|---|
| 227 | Parameters.Add(new ValueLookupParameter<ISymbolicExpressionTreeInterpreter>(SymbolicExpressionTreeInterpreterParameterName, "The interpreter that should be used for the analysis of symbolic expression trees."));
|
|---|
| 228 | Parameters.Add(new LookupParameter<ClassificationProblemData>(ClassificationProblemDataParameterName, "The problem data for which the symbolic expression tree is a solution."));
|
|---|
| 229 | Parameters.Add(new LookupParameter<ISymbolicClassificationEvaluator>(EvaluatorParameterName, "The evaluator which should be used to evaluate the solution on the validation set."));
|
|---|
| 230 | Parameters.Add(new ValueLookupParameter<IntValue>(ValidationSamplesStartParameterName, "The first index of the validation partition of the data set."));
|
|---|
| 231 | Parameters.Add(new ValueLookupParameter<IntValue>(ValidationSamplesEndParameterName, "The last index of the validation partition of the data set."));
|
|---|
| 232 | Parameters.Add(new ValueParameter<PercentValue>(RelativeNumberOfEvaluatedSamplesParameterName, "The relative number of samples of the dataset partition, which should be randomly chosen for evaluation between the start and end index.", new PercentValue(1)));
|
|---|
| 233 | Parameters.Add(new ValueLookupParameter<DoubleValue>(UpperEstimationLimitParameterName, "The upper estimation limit that was set for the evaluation of the symbolic expression trees."));
|
|---|
| 234 | Parameters.Add(new ValueLookupParameter<DoubleValue>(LowerEstimationLimitParameterName, "The lower estimation limit that was set for the evaluation of the symbolic expression trees."));
|
|---|
| 235 | Parameters.Add(new LookupParameter<DataTable>(VariableFrequenciesParameterName, "The variable frequencies table to use for the calculation of variable impacts"));
|
|---|
| 236 | Parameters.Add(new ValueParameter<BoolValue>(CalculateSolutionComplexityParameterName, "Determines if the length and height of the validation best solution should be calculated.", new BoolValue(true)));
|
|---|
| 237 | Parameters.Add(new ValueLookupParameter<BoolValue>(ApplyLinearScalingParameterName, "The switch determines if the best solution should be linearly scaled on the whole training set.", new BoolValue(false)));
|
|---|
| 238 |
|
|---|
| 239 | Parameters.Add(new ValueLookupParameter<ResultCollection>(ResultsParameterName, "The results collection where the analysis values should be stored."));
|
|---|
| 240 | Parameters.Add(new LookupParameter<DoubleValue>(BestValidationQualityParameterName, "The validation quality of the best solution in the current run."));
|
|---|
| 241 | Parameters.Add(new LookupParameter<SymbolicClassificationSolution>(BestValidationSolutionParameterName, "The best solution on the validation data found in the current run."));
|
|---|
| 242 | Parameters.Add(new LookupParameter<DoubleValue>(BestSolutionAccuracyTrainingParameterName, "The training accuracy of the best solution."));
|
|---|
| 243 | Parameters.Add(new LookupParameter<DoubleValue>(BestSolutionAccuracyTestParameterName, "The test accuracy of the best solution."));
|
|---|
| 244 | Parameters.Add(new LookupParameter<IntValue>(BestSolutionLengthParameterName, "The length of the best symbolic classification solution."));
|
|---|
| 245 | Parameters.Add(new LookupParameter<IntValue>(BestSolutionHeightParameterName, "The height of the best symbolic classification solution."));
|
|---|
| 246 | }
|
|---|
| 247 |
|
|---|
| 248 | [StorableHook(HookType.AfterDeserialization)]
|
|---|
| 249 | private void AfterDeserialization() {
|
|---|
| 250 | if (!Parameters.ContainsKey(CalculateSolutionComplexityParameterName)) {
|
|---|
| 251 | Parameters.Add(new ValueParameter<BoolValue>(CalculateSolutionComplexityParameterName, "Determines if the length and height of the validation best solution should be calculated.", new BoolValue(true)));
|
|---|
| 252 | }
|
|---|
| 253 | if (!Parameters.ContainsKey(BestSolutionLengthParameterName)) {
|
|---|
| 254 | Parameters.Add(new LookupParameter<IntValue>(BestSolutionLengthParameterName, "The length of the best symbolic classification solution."));
|
|---|
| 255 | }
|
|---|
| 256 | if (!Parameters.ContainsKey(BestSolutionHeightParameterName)) {
|
|---|
| 257 | Parameters.Add(new LookupParameter<IntValue>(BestSolutionHeightParameterName, "The height of the best symbolic classification solution."));
|
|---|
| 258 | }
|
|---|
| 259 | if (!Parameters.ContainsKey(ApplyLinearScalingParameterName)) {
|
|---|
| 260 | Parameters.Add(new ValueLookupParameter<BoolValue>(ApplyLinearScalingParameterName, "The switch determines if the best solution should be linearly scaled on the whole training set.", new BoolValue(false)));
|
|---|
| 261 | }
|
|---|
| 262 | }
|
|---|
| 263 |
|
|---|
| 264 | public override IDeepCloneable Clone(Cloner cloner) {
|
|---|
| 265 | return new ValidationBestSymbolicClassificationSolutionAnalyzer(this, cloner);
|
|---|
| 266 | }
|
|---|
| 267 |
|
|---|
| 268 | public override IOperation Apply() {
|
|---|
| 269 | var trees = SymbolicExpressionTree;
|
|---|
| 270 | string targetVariable = ClassificationProblemData.TargetVariable.Value;
|
|---|
| 271 |
|
|---|
| 272 | // select a random subset of rows in the validation set
|
|---|
| 273 | int validationStart = ValidiationSamplesStart.Value;
|
|---|
| 274 | int validationEnd = ValidationSamplesEnd.Value;
|
|---|
| 275 | int seed = Random.Next();
|
|---|
| 276 | int count = (int)((validationEnd - validationStart) * RelativeNumberOfEvaluatedSamples.Value);
|
|---|
| 277 | if (count == 0) count = 1;
|
|---|
| 278 | IEnumerable<int> rows = RandomEnumerable.SampleRandomNumbers(seed, validationStart, validationEnd, count)
|
|---|
| 279 | .Where(row => row < ClassificationProblemData.TestSamplesStart.Value || ClassificationProblemData.TestSamplesEnd.Value <= row);
|
|---|
| 280 |
|
|---|
| 281 | double upperEstimationLimit = UpperEstimationLimit != null ? UpperEstimationLimit.Value : double.PositiveInfinity;
|
|---|
| 282 | double lowerEstimationLimit = LowerEstimationLimit != null ? LowerEstimationLimit.Value : double.NegativeInfinity;
|
|---|
| 283 |
|
|---|
| 284 | double bestQuality = Maximization.Value ? double.NegativeInfinity : double.PositiveInfinity;
|
|---|
| 285 | SymbolicExpressionTree bestTree = null;
|
|---|
| 286 |
|
|---|
| 287 | foreach (var tree in trees) {
|
|---|
| 288 | double quality = Evaluator.Evaluate(SymbolicExpressionTreeInterpreter, tree,
|
|---|
| 289 | lowerEstimationLimit, upperEstimationLimit, ClassificationProblemData.Dataset,
|
|---|
| 290 | targetVariable, rows);
|
|---|
| 291 |
|
|---|
| 292 | if ((Maximization.Value && quality > bestQuality) ||
|
|---|
| 293 | (!Maximization.Value && quality < bestQuality)) {
|
|---|
| 294 | bestQuality = quality;
|
|---|
| 295 | bestTree = tree;
|
|---|
| 296 | }
|
|---|
| 297 | }
|
|---|
| 298 |
|
|---|
| 299 | // if the best validation tree is better than the current best solution => update
|
|---|
| 300 | bool newBest =
|
|---|
| 301 | BestValidationQuality == null ||
|
|---|
| 302 | (Maximization.Value && bestQuality > BestValidationQuality.Value) ||
|
|---|
| 303 | (!Maximization.Value && bestQuality < BestValidationQuality.Value);
|
|---|
| 304 | if (newBest) {
|
|---|
| 305 | if (ApplyLinearScaling.Value) {
|
|---|
| 306 | double alpha, beta;
|
|---|
| 307 | SymbolicRegressionScaledMeanSquaredErrorEvaluator.Calculate(SymbolicExpressionTreeInterpreter, bestTree,
|
|---|
| 308 | lowerEstimationLimit, upperEstimationLimit,
|
|---|
| 309 | ClassificationProblemData.Dataset, targetVariable,
|
|---|
| 310 | ClassificationProblemData.TrainingIndizes, out beta, out alpha);
|
|---|
| 311 |
|
|---|
| 312 | // scale tree for solution
|
|---|
| 313 | bestTree = SymbolicRegressionSolutionLinearScaler.Scale(bestTree, alpha, beta);
|
|---|
| 314 | }
|
|---|
| 315 | var model = new SymbolicRegressionModel((ISymbolicExpressionTreeInterpreter)SymbolicExpressionTreeInterpreter.Clone(),
|
|---|
| 316 | bestTree);
|
|---|
| 317 |
|
|---|
| 318 | if (BestValidationSolution == null) {
|
|---|
| 319 | BestValidationSolution = new SymbolicClassificationSolution(ClassificationProblemData, model, LowerEstimationLimit.Value, UpperEstimationLimit.Value);
|
|---|
| 320 | BestValidationSolution.Name = BestValidationSolutionParameterName;
|
|---|
| 321 | BestValidationSolution.Description = "Best solution on validation partition found over the whole run.";
|
|---|
| 322 | BestValidationQuality = new DoubleValue(bestQuality);
|
|---|
| 323 | } else {
|
|---|
| 324 | BestValidationSolution.Model = model;
|
|---|
| 325 | BestValidationQuality.Value = bestQuality;
|
|---|
| 326 | }
|
|---|
| 327 |
|
|---|
| 328 | UpdateBestSolutionResults();
|
|---|
| 329 | }
|
|---|
| 330 | return base.Apply();
|
|---|
| 331 | }
|
|---|
| 332 |
|
|---|
| 333 | private void UpdateBestSolutionResults() {
|
|---|
| 334 | if (CalculateSolutionComplexity.Value) {
|
|---|
| 335 | BestSolutionLength = new IntValue(BestValidationSolution.Model.SymbolicExpressionTree.Size);
|
|---|
| 336 | BestSolutionHeight = new IntValue(BestValidationSolution.Model.SymbolicExpressionTree.Height);
|
|---|
| 337 | if (!Results.ContainsKey(BestSolutionLengthParameterName)) {
|
|---|
| 338 | Results.Add(new Result(BestSolutionLengthParameterName, "Length of the best solution on the validation set", new IntValue()));
|
|---|
| 339 | Results.Add(new Result(BestSolutionHeightParameterName, "Height of the best solution on the validation set", new IntValue()));
|
|---|
| 340 | }
|
|---|
| 341 | Results[BestSolutionLengthParameterName].Value = BestSolutionLength;
|
|---|
| 342 | Results[BestSolutionHeightParameterName].Value = BestSolutionHeight;
|
|---|
| 343 | }
|
|---|
| 344 |
|
|---|
| 345 | BestSymbolicRegressionSolutionAnalyzer.UpdateBestSolutionResults(BestValidationSolution, ClassificationProblemData, Results, Generations, VariableFrequencies);
|
|---|
| 346 |
|
|---|
| 347 | IEnumerable<double> trainingValues = ClassificationProblemData.Dataset.GetEnumeratedVariableValues(
|
|---|
| 348 | ClassificationProblemData.TargetVariable.Value, ClassificationProblemData.TrainingIndizes);
|
|---|
| 349 | IEnumerable<double> testValues = ClassificationProblemData.Dataset.GetEnumeratedVariableValues(
|
|---|
| 350 | ClassificationProblemData.TargetVariable.Value, ClassificationProblemData.TestIndizes);
|
|---|
| 351 |
|
|---|
| 352 | OnlineAccuracyEvaluator accuracyEvaluator = new OnlineAccuracyEvaluator();
|
|---|
| 353 | var originalEnumerator = trainingValues.GetEnumerator();
|
|---|
| 354 | var estimatedEnumerator = BestValidationSolution.EstimatedTrainingClassValues.GetEnumerator();
|
|---|
| 355 | while (originalEnumerator.MoveNext() & estimatedEnumerator.MoveNext()) {
|
|---|
| 356 | accuracyEvaluator.Add(originalEnumerator.Current, estimatedEnumerator.Current);
|
|---|
| 357 | }
|
|---|
| 358 | double trainingAccuracy = accuracyEvaluator.Accuracy;
|
|---|
| 359 |
|
|---|
| 360 | accuracyEvaluator.Reset();
|
|---|
| 361 | originalEnumerator = testValues.GetEnumerator();
|
|---|
| 362 | estimatedEnumerator = BestValidationSolution.EstimatedTestClassValues.GetEnumerator();
|
|---|
| 363 | while (originalEnumerator.MoveNext() & estimatedEnumerator.MoveNext()) {
|
|---|
| 364 | accuracyEvaluator.Add(originalEnumerator.Current, estimatedEnumerator.Current);
|
|---|
| 365 | }
|
|---|
| 366 | double testAccuracy = accuracyEvaluator.Accuracy;
|
|---|
| 367 |
|
|---|
| 368 | if (!Results.ContainsKey(BestSolutionAccuracyTrainingParameterName)) {
|
|---|
| 369 | BestSolutionAccuracyTraining = new DoubleValue(trainingAccuracy);
|
|---|
| 370 | BestSolutionAccuracyTest = new DoubleValue(testAccuracy);
|
|---|
| 371 |
|
|---|
| 372 | Results.Add(new Result(BestSolutionAccuracyTrainingParameterName, BestSolutionAccuracyTraining));
|
|---|
| 373 | Results.Add(new Result(BestSolutionAccuracyTestParameterName, BestSolutionAccuracyTest));
|
|---|
| 374 | } else {
|
|---|
| 375 | BestSolutionAccuracyTraining.Value = trainingAccuracy;
|
|---|
| 376 | BestSolutionAccuracyTest.Value = testAccuracy;
|
|---|
| 377 | }
|
|---|
| 378 | }
|
|---|
| 379 | }
|
|---|
| 380 | }
|
|---|
