[10072] | 1 | #region License Information
|
---|
| 2 | /* HeuristicLab
|
---|
[12012] | 3 | * Copyright (C) 2002-2015 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
|
---|
[10072] | 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/>.
|
---|
[10968] | 19 | *
|
---|
| 20 | * Author: Sabine Winkler
|
---|
[10072] | 21 | */
|
---|
[10968] | 22 |
|
---|
[10072] | 23 | #endregion
|
---|
| 24 |
|
---|
| 25 | using System.Linq;
|
---|
| 26 | using HeuristicLab.Common;
|
---|
| 27 | using HeuristicLab.Core;
|
---|
[10073] | 28 | using HeuristicLab.Encodings.IntegerVectorEncoding;
|
---|
[10072] | 29 | using HeuristicLab.Parameters;
|
---|
| 30 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
---|
[10073] | 31 | using HeuristicLab.Problems.DataAnalysis;
|
---|
| 32 | using HeuristicLab.Problems.DataAnalysis.Symbolic;
|
---|
| 33 | using HeuristicLab.Problems.DataAnalysis.Symbolic.Regression;
|
---|
[10072] | 34 |
|
---|
[10073] | 35 | namespace HeuristicLab.Problems.GrammaticalEvolution {
|
---|
[10276] | 36 | [Item("Grammatical Evolution Symbolic Regression Problem (single objective)",
|
---|
[10974] | 37 | "Represents grammatical evolution for single objective symbolic regression problems.")]
|
---|
[10072] | 38 | [StorableClass]
|
---|
[12504] | 39 | [Creatable(CreatableAttribute.Categories.GeneticProgrammingProblems, Priority = 180)]
|
---|
[10075] | 40 | public class GESymbolicRegressionSingleObjectiveProblem : GESymbolicDataAnalysisSingleObjectiveProblem<IRegressionProblemData, IGESymbolicRegressionSingleObjectiveEvaluator, IIntegerVectorCreator>,
|
---|
| 41 | IRegressionProblem {
|
---|
[10072] | 42 | private const double PunishmentFactor = 10;
|
---|
[10328] | 43 | private const int InitialMaximumTreeLength = 30;
|
---|
[10072] | 44 | private const string EstimationLimitsParameterName = "EstimationLimits";
|
---|
[10276] | 45 | private const string EstimationLimitsParameterDescription
|
---|
| 46 | = "The limits for the estimated value that can be returned by the symbolic regression model.";
|
---|
[10072] | 47 |
|
---|
| 48 | #region parameter properties
|
---|
| 49 | public IFixedValueParameter<DoubleLimit> EstimationLimitsParameter {
|
---|
| 50 | get { return (IFixedValueParameter<DoubleLimit>)Parameters[EstimationLimitsParameterName]; }
|
---|
| 51 | }
|
---|
| 52 | #endregion
|
---|
| 53 | #region properties
|
---|
| 54 | public DoubleLimit EstimationLimits {
|
---|
| 55 | get { return EstimationLimitsParameter.Value; }
|
---|
| 56 | }
|
---|
| 57 | #endregion
|
---|
| 58 | [StorableConstructor]
|
---|
[10073] | 59 | protected GESymbolicRegressionSingleObjectiveProblem(bool deserializing) : base(deserializing) { }
|
---|
| 60 | protected GESymbolicRegressionSingleObjectiveProblem(GESymbolicRegressionSingleObjectiveProblem original, Cloner cloner)
|
---|
[10072] | 61 | : base(original, cloner) {
|
---|
| 62 | RegisterEventHandlers();
|
---|
| 63 | }
|
---|
[10073] | 64 | public override IDeepCloneable Clone(Cloner cloner) { return new GESymbolicRegressionSingleObjectiveProblem(this, cloner); }
|
---|
[10072] | 65 |
|
---|
[10073] | 66 | public GESymbolicRegressionSingleObjectiveProblem()
|
---|
[10263] | 67 | : base(new RegressionProblemData(), new GESymbolicRegressionSingleObjectiveEvaluator(), new UniformRandomIntegerVectorCreator()) {
|
---|
[10072] | 68 | Parameters.Add(new FixedValueParameter<DoubleLimit>(EstimationLimitsParameterName, EstimationLimitsParameterDescription));
|
---|
| 69 |
|
---|
| 70 | EstimationLimitsParameter.Hidden = true;
|
---|
| 71 |
|
---|
| 72 |
|
---|
| 73 | ApplyLinearScalingParameter.Value.Value = true;
|
---|
[10974] | 74 | Maximization.Value = Evaluator.Maximization;
|
---|
[10072] | 75 | MaximumSymbolicExpressionTreeLength.Value = InitialMaximumTreeLength;
|
---|
| 76 |
|
---|
| 77 | RegisterEventHandlers();
|
---|
| 78 | InitializeOperators();
|
---|
| 79 | UpdateEstimationLimits();
|
---|
| 80 | }
|
---|
| 81 |
|
---|
| 82 | [StorableHook(HookType.AfterDeserialization)]
|
---|
| 83 | private void AfterDeserialization() {
|
---|
| 84 | RegisterEventHandlers();
|
---|
| 85 | }
|
---|
| 86 |
|
---|
| 87 | private void RegisterEventHandlers() {
|
---|
[10974] | 88 | // when the ge evaluator itself changes
|
---|
| 89 | EvaluatorParameter.ValueChanged += (sender, args) => {
|
---|
| 90 | // register a new hander for the symbreg evaluator in the ge evaluator
|
---|
| 91 | // hacky because we the evaluator does not have an event for changes of the maximization property
|
---|
| 92 | EvaluatorParameter.Value.EvaluatorParameter.ValueChanged +=
|
---|
| 93 | (_, __) => Maximization.Value = Evaluator.Maximization;
|
---|
| 94 | };
|
---|
| 95 | EvaluatorParameter.Value.EvaluatorParameter.ValueChanged +=
|
---|
| 96 | (sender, args) => Maximization.Value = Evaluator.Maximization;
|
---|
[10072] | 97 | }
|
---|
| 98 |
|
---|
| 99 |
|
---|
| 100 | private void InitializeOperators() {
|
---|
| 101 | Operators.Add(new SymbolicRegressionSingleObjectiveTrainingBestSolutionAnalyzer());
|
---|
| 102 | Operators.Add(new SymbolicRegressionSingleObjectiveValidationBestSolutionAnalyzer());
|
---|
| 103 | Operators.Add(new SymbolicRegressionSingleObjectiveOverfittingAnalyzer());
|
---|
| 104 | Operators.Add(new SymbolicRegressionSingleObjectiveTrainingParetoBestSolutionAnalyzer());
|
---|
| 105 | Operators.Add(new SymbolicRegressionSingleObjectiveValidationParetoBestSolutionAnalyzer());
|
---|
| 106 |
|
---|
| 107 | ParameterizeOperators();
|
---|
| 108 | }
|
---|
| 109 |
|
---|
| 110 | private void UpdateEstimationLimits() {
|
---|
| 111 | if (ProblemData.TrainingIndices.Any()) {
|
---|
| 112 | var targetValues = ProblemData.Dataset.GetDoubleValues(ProblemData.TargetVariable, ProblemData.TrainingIndices).ToList();
|
---|
| 113 | var mean = targetValues.Average();
|
---|
| 114 | var range = targetValues.Max() - targetValues.Min();
|
---|
| 115 | EstimationLimits.Upper = mean + PunishmentFactor * range;
|
---|
| 116 | EstimationLimits.Lower = mean - PunishmentFactor * range;
|
---|
| 117 | } else {
|
---|
| 118 | EstimationLimits.Upper = double.MaxValue;
|
---|
| 119 | EstimationLimits.Lower = double.MinValue;
|
---|
| 120 | }
|
---|
| 121 | }
|
---|
| 122 |
|
---|
| 123 | protected override void OnProblemDataChanged() {
|
---|
| 124 | base.OnProblemDataChanged();
|
---|
| 125 | UpdateEstimationLimits();
|
---|
| 126 | }
|
---|
| 127 |
|
---|
| 128 | protected override void ParameterizeOperators() {
|
---|
| 129 | base.ParameterizeOperators();
|
---|
| 130 | if (Parameters.ContainsKey(EstimationLimitsParameterName)) {
|
---|
| 131 | var operators = Parameters.OfType<IValueParameter>().Select(p => p.Value).OfType<IOperator>().Union(Operators);
|
---|
| 132 | foreach (var op in operators.OfType<ISymbolicDataAnalysisBoundedOperator>()) {
|
---|
| 133 | op.EstimationLimitsParameter.ActualName = EstimationLimitsParameter.Name;
|
---|
| 134 | }
|
---|
| 135 | }
|
---|
| 136 | }
|
---|
| 137 | }
|
---|
| 138 | }
|
---|