[16927] | 1 | #region License Information
|
---|
[5607] | 2 | /* HeuristicLab
|
---|
[17206] | 3 | * Copyright (C) Heuristic and Evolutionary Algorithms Laboratory (HEAL)
|
---|
[5607] | 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 |
|
---|
[8723] | 22 | using System.Linq;
|
---|
[16713] | 23 | using HEAL.Attic;
|
---|
[5607] | 24 | using HeuristicLab.Common;
|
---|
| 25 | using HeuristicLab.Core;
|
---|
| 26 | using HeuristicLab.Data;
|
---|
[11332] | 27 | using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
|
---|
[5914] | 28 | using HeuristicLab.Optimization;
|
---|
[5607] | 29 |
|
---|
[5624] | 30 | namespace HeuristicLab.Problems.DataAnalysis.Symbolic.Regression {
|
---|
[5607] | 31 | /// <summary>
|
---|
| 32 | /// Represents a symbolic regression solution (model + data) and attributes of the solution like accuracy and complexity
|
---|
| 33 | /// </summary>
|
---|
[16641] | 34 | [StorableType("88E56AF9-AD72-47E4-A613-8875703BD927")]
|
---|
[5607] | 35 | [Item(Name = "SymbolicRegressionSolution", Description = "Represents a symbolic regression solution (model + data) and attributes of the solution like accuracy and complexity.")]
|
---|
[5717] | 36 | public sealed class SymbolicRegressionSolution : RegressionSolution, ISymbolicRegressionSolution {
|
---|
[5975] | 37 | private const string ModelLengthResultName = "Model Length";
|
---|
| 38 | private const string ModelDepthResultName = "Model Depth";
|
---|
[5736] | 39 |
|
---|
[8723] | 40 | private const string EstimationLimitsResultsResultName = "Estimation Limits Results";
|
---|
| 41 | private const string EstimationLimitsResultName = "Estimation Limits";
|
---|
| 42 | private const string TrainingUpperEstimationLimitHitsResultName = "Training Upper Estimation Limit Hits";
|
---|
| 43 | private const string TestLowerEstimationLimitHitsResultName = "Test Lower Estimation Limit Hits";
|
---|
| 44 | private const string TrainingLowerEstimationLimitHitsResultName = "Training Lower Estimation Limit Hits";
|
---|
| 45 | private const string TestUpperEstimationLimitHitsResultName = "Test Upper Estimation Limit Hits";
|
---|
| 46 | private const string TrainingNaNEvaluationsResultName = "Training NaN Evaluations";
|
---|
| 47 | private const string TestNaNEvaluationsResultName = "Test NaN Evaluations";
|
---|
| 48 |
|
---|
[17509] | 49 | private const string ModelBoundsResultName = "Model Bounds";
|
---|
[16556] | 50 |
|
---|
[5624] | 51 | public new ISymbolicRegressionModel Model {
|
---|
| 52 | get { return (ISymbolicRegressionModel)base.Model; }
|
---|
[5717] | 53 | set { base.Model = value; }
|
---|
[5607] | 54 | }
|
---|
[5624] | 55 | ISymbolicDataAnalysisModel ISymbolicDataAnalysisSolution.Model {
|
---|
| 56 | get { return (ISymbolicDataAnalysisModel)base.Model; }
|
---|
[5607] | 57 | }
|
---|
[5736] | 58 | public int ModelLength {
|
---|
| 59 | get { return ((IntValue)this[ModelLengthResultName].Value).Value; }
|
---|
| 60 | private set { ((IntValue)this[ModelLengthResultName].Value).Value = value; }
|
---|
| 61 | }
|
---|
[5607] | 62 |
|
---|
[5736] | 63 | public int ModelDepth {
|
---|
| 64 | get { return ((IntValue)this[ModelDepthResultName].Value).Value; }
|
---|
| 65 | private set { ((IntValue)this[ModelDepthResultName].Value).Value = value; }
|
---|
| 66 | }
|
---|
| 67 |
|
---|
[8723] | 68 | private ResultCollection EstimationLimitsResultCollection {
|
---|
| 69 | get { return (ResultCollection)this[EstimationLimitsResultsResultName].Value; }
|
---|
| 70 | }
|
---|
| 71 | public DoubleLimit EstimationLimits {
|
---|
| 72 | get { return (DoubleLimit)EstimationLimitsResultCollection[EstimationLimitsResultName].Value; }
|
---|
| 73 | }
|
---|
| 74 |
|
---|
| 75 | public int TrainingUpperEstimationLimitHits {
|
---|
| 76 | get { return ((IntValue)EstimationLimitsResultCollection[TrainingUpperEstimationLimitHitsResultName].Value).Value; }
|
---|
| 77 | private set { ((IntValue)EstimationLimitsResultCollection[TrainingUpperEstimationLimitHitsResultName].Value).Value = value; }
|
---|
| 78 | }
|
---|
| 79 | public int TestUpperEstimationLimitHits {
|
---|
| 80 | get { return ((IntValue)EstimationLimitsResultCollection[TestUpperEstimationLimitHitsResultName].Value).Value; }
|
---|
| 81 | private set { ((IntValue)EstimationLimitsResultCollection[TestUpperEstimationLimitHitsResultName].Value).Value = value; }
|
---|
| 82 | }
|
---|
| 83 | public int TrainingLowerEstimationLimitHits {
|
---|
| 84 | get { return ((IntValue)EstimationLimitsResultCollection[TrainingLowerEstimationLimitHitsResultName].Value).Value; }
|
---|
| 85 | private set { ((IntValue)EstimationLimitsResultCollection[TrainingLowerEstimationLimitHitsResultName].Value).Value = value; }
|
---|
| 86 | }
|
---|
| 87 | public int TestLowerEstimationLimitHits {
|
---|
| 88 | get { return ((IntValue)EstimationLimitsResultCollection[TestLowerEstimationLimitHitsResultName].Value).Value; }
|
---|
| 89 | private set { ((IntValue)EstimationLimitsResultCollection[TestLowerEstimationLimitHitsResultName].Value).Value = value; }
|
---|
| 90 | }
|
---|
| 91 | public int TrainingNaNEvaluations {
|
---|
| 92 | get { return ((IntValue)EstimationLimitsResultCollection[TrainingNaNEvaluationsResultName].Value).Value; }
|
---|
| 93 | private set { ((IntValue)EstimationLimitsResultCollection[TrainingNaNEvaluationsResultName].Value).Value = value; }
|
---|
| 94 | }
|
---|
| 95 | public int TestNaNEvaluations {
|
---|
| 96 | get { return ((IntValue)EstimationLimitsResultCollection[TestNaNEvaluationsResultName].Value).Value; }
|
---|
| 97 | private set { ((IntValue)EstimationLimitsResultCollection[TestNaNEvaluationsResultName].Value).Value = value; }
|
---|
| 98 | }
|
---|
| 99 |
|
---|
[17509] | 100 | public IntervalCollection ModelBoundsCollection {
|
---|
[17555] | 101 | get {
|
---|
| 102 | if (!ContainsKey(ModelBoundsResultName)) return null;
|
---|
| 103 | return (IntervalCollection)this[ModelBoundsResultName].Value;
|
---|
| 104 | }
|
---|
[17553] | 105 | private set {
|
---|
[17555] | 106 | if (ContainsKey(ModelBoundsResultName)) {
|
---|
[17553] | 107 | this[ModelBoundsResultName].Value = value;
|
---|
[17555] | 108 | } else {
|
---|
| 109 | Add(new Result(ModelBoundsResultName, "Results concerning the derivation of symbolic regression solution", value));
|
---|
| 110 | }
|
---|
| 111 |
|
---|
[17553] | 112 | }
|
---|
[16627] | 113 | }
|
---|
[16556] | 114 |
|
---|
| 115 |
|
---|
[16627] | 116 |
|
---|
[5607] | 117 | [StorableConstructor]
|
---|
[16627] | 118 | private SymbolicRegressionSolution(StorableConstructorFlag _) : base(_) { }
|
---|
[5717] | 119 | private SymbolicRegressionSolution(SymbolicRegressionSolution original, Cloner cloner)
|
---|
[5607] | 120 | : base(original, cloner) {
|
---|
| 121 | }
|
---|
[5624] | 122 | public SymbolicRegressionSolution(ISymbolicRegressionModel model, IRegressionProblemData problemData)
|
---|
| 123 | : base(model, problemData) {
|
---|
[11332] | 124 | foreach (var node in model.SymbolicExpressionTree.Root.IterateNodesPrefix().OfType<SymbolicExpressionTreeTopLevelNode>())
|
---|
| 125 | node.SetGrammar(null);
|
---|
| 126 |
|
---|
[5736] | 127 | Add(new Result(ModelLengthResultName, "Length of the symbolic regression model.", new IntValue()));
|
---|
| 128 | Add(new Result(ModelDepthResultName, "Depth of the symbolic regression model.", new IntValue()));
|
---|
[8723] | 129 |
|
---|
| 130 | ResultCollection estimationLimitResults = new ResultCollection();
|
---|
| 131 | estimationLimitResults.Add(new Result(EstimationLimitsResultName, "", new DoubleLimit()));
|
---|
| 132 | estimationLimitResults.Add(new Result(TrainingUpperEstimationLimitHitsResultName, "", new IntValue()));
|
---|
| 133 | estimationLimitResults.Add(new Result(TestUpperEstimationLimitHitsResultName, "", new IntValue()));
|
---|
| 134 | estimationLimitResults.Add(new Result(TrainingLowerEstimationLimitHitsResultName, "", new IntValue()));
|
---|
| 135 | estimationLimitResults.Add(new Result(TestLowerEstimationLimitHitsResultName, "", new IntValue()));
|
---|
| 136 | estimationLimitResults.Add(new Result(TrainingNaNEvaluationsResultName, "", new IntValue()));
|
---|
| 137 | estimationLimitResults.Add(new Result(TestNaNEvaluationsResultName, "", new IntValue()));
|
---|
| 138 | Add(new Result(EstimationLimitsResultsResultName, "Results concerning the estimation limits of symbolic regression solution", estimationLimitResults));
|
---|
[17538] | 139 |
|
---|
| 140 | if (IntervalInterpreter.IsCompatible(Model.SymbolicExpressionTree))
|
---|
| 141 | Add(new Result(ModelBoundsResultName, "Results concerning the derivation of symbolic regression solution", new IntervalCollection()));
|
---|
| 142 |
|
---|
[6588] | 143 | RecalculateResults();
|
---|
[5607] | 144 | }
|
---|
| 145 |
|
---|
| 146 | public override IDeepCloneable Clone(Cloner cloner) {
|
---|
| 147 | return new SymbolicRegressionSolution(this, cloner);
|
---|
| 148 | }
|
---|
[5729] | 149 |
|
---|
[8723] | 150 | [StorableHook(HookType.AfterDeserialization)]
|
---|
| 151 | private void AfterDeserialization() {
|
---|
| 152 | if (!ContainsKey(EstimationLimitsResultsResultName)) {
|
---|
| 153 | ResultCollection estimationLimitResults = new ResultCollection();
|
---|
| 154 | estimationLimitResults.Add(new Result(EstimationLimitsResultName, "", new DoubleLimit()));
|
---|
| 155 | estimationLimitResults.Add(new Result(TrainingUpperEstimationLimitHitsResultName, "", new IntValue()));
|
---|
| 156 | estimationLimitResults.Add(new Result(TestUpperEstimationLimitHitsResultName, "", new IntValue()));
|
---|
| 157 | estimationLimitResults.Add(new Result(TrainingLowerEstimationLimitHitsResultName, "", new IntValue()));
|
---|
| 158 | estimationLimitResults.Add(new Result(TestLowerEstimationLimitHitsResultName, "", new IntValue()));
|
---|
| 159 | estimationLimitResults.Add(new Result(TrainingNaNEvaluationsResultName, "", new IntValue()));
|
---|
| 160 | estimationLimitResults.Add(new Result(TestNaNEvaluationsResultName, "", new IntValue()));
|
---|
| 161 | Add(new Result(EstimationLimitsResultsResultName, "Results concerning the estimation limits of symbolic regression solution", estimationLimitResults));
|
---|
| 162 | CalculateResults();
|
---|
| 163 | }
|
---|
[17509] | 164 |
|
---|
| 165 | if (!ContainsKey(ModelBoundsResultName)) {
|
---|
[17553] | 166 | if (IntervalInterpreter.IsCompatible(Model.SymbolicExpressionTree)) {
|
---|
| 167 | Add(new Result(ModelBoundsResultName, "Results concerning the derivation of symbolic regression solution", new IntervalCollection()));
|
---|
| 168 | CalculateResults();
|
---|
| 169 | }
|
---|
[17509] | 170 | }
|
---|
[8723] | 171 | }
|
---|
| 172 |
|
---|
[6411] | 173 | protected override void RecalculateResults() {
|
---|
[6602] | 174 | base.RecalculateResults();
|
---|
[8723] | 175 | CalculateResults();
|
---|
| 176 | }
|
---|
| 177 |
|
---|
| 178 | private void CalculateResults() {
|
---|
[5736] | 179 | ModelLength = Model.SymbolicExpressionTree.Length;
|
---|
| 180 | ModelDepth = Model.SymbolicExpressionTree.Depth;
|
---|
[8723] | 181 |
|
---|
| 182 | EstimationLimits.Lower = Model.LowerEstimationLimit;
|
---|
| 183 | EstimationLimits.Upper = Model.UpperEstimationLimit;
|
---|
| 184 |
|
---|
| 185 | TrainingUpperEstimationLimitHits = EstimatedTrainingValues.Count(x => x.IsAlmost(Model.UpperEstimationLimit));
|
---|
| 186 | TestUpperEstimationLimitHits = EstimatedTestValues.Count(x => x.IsAlmost(Model.UpperEstimationLimit));
|
---|
| 187 | TrainingLowerEstimationLimitHits = EstimatedTrainingValues.Count(x => x.IsAlmost(Model.LowerEstimationLimit));
|
---|
| 188 | TestLowerEstimationLimitHits = EstimatedTestValues.Count(x => x.IsAlmost(Model.LowerEstimationLimit));
|
---|
| 189 | TrainingNaNEvaluations = Model.Interpreter.GetSymbolicExpressionTreeValues(Model.SymbolicExpressionTree, ProblemData.Dataset, ProblemData.TrainingIndices).Count(double.IsNaN);
|
---|
| 190 | TestNaNEvaluations = Model.Interpreter.GetSymbolicExpressionTreeValues(Model.SymbolicExpressionTree, ProblemData.Dataset, ProblemData.TestIndices).Count(double.IsNaN);
|
---|
[16627] | 191 |
|
---|
[17509] | 192 | //Check if the tree contains unknown symbols for the interval calculation
|
---|
[17538] | 193 | if (IntervalInterpreter.IsCompatible(Model.SymbolicExpressionTree))
|
---|
[17509] | 194 | ModelBoundsCollection = CalculateModelIntervals(this);
|
---|
[5736] | 195 | }
|
---|
[16851] | 196 |
|
---|
[16904] | 197 | private static IntervalCollection CalculateModelIntervals(ISymbolicRegressionSolution solution) {
|
---|
| 198 | var intervalEvaluation = new IntervalCollection();
|
---|
[16851] | 199 | var interpreter = new IntervalInterpreter();
|
---|
| 200 | var problemData = solution.ProblemData;
|
---|
| 201 | var model = solution.Model;
|
---|
[17564] | 202 | var variableRanges = problemData.VariableRanges.GetReadonlyDictionary();
|
---|
[16851] | 203 |
|
---|
[16927] | 204 | intervalEvaluation.AddInterval($"Target {problemData.TargetVariable}", new Interval(variableRanges[problemData.TargetVariable].LowerBound, variableRanges[problemData.TargetVariable].UpperBound));
|
---|
| 205 | intervalEvaluation.AddInterval("Model Interval", interpreter.GetSymbolicExpressionTreeInterval(model.SymbolicExpressionTree, variableRanges));
|
---|
[16851] | 206 |
|
---|
[17538] | 207 | if (DerivativeCalculator.IsCompatible(model.SymbolicExpressionTree)) {
|
---|
| 208 | foreach (var inputVariable in model.VariablesUsedForPrediction.OrderBy(v => v, new NaturalStringComparer())) {
|
---|
| 209 | var derivedModel = DerivativeCalculator.Derive(model.SymbolicExpressionTree, inputVariable);
|
---|
| 210 | var derivedResultInterval = interpreter.GetSymbolicExpressionTreeInterval(derivedModel, variableRanges);
|
---|
| 211 |
|
---|
| 212 | intervalEvaluation.AddInterval(" ∂f/∂" + inputVariable, new Interval(derivedResultInterval.LowerBound, derivedResultInterval.UpperBound));
|
---|
| 213 | }
|
---|
[16851] | 214 | }
|
---|
| 215 |
|
---|
| 216 | return intervalEvaluation;
|
---|
| 217 | }
|
---|
[5607] | 218 | }
|
---|
| 219 | }
|
---|