| 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.Common;
|
|---|
| 25 | using HeuristicLab.Core;
|
|---|
| 26 | using HeuristicLab.Data;
|
|---|
| 27 | using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
|
|---|
| 28 | using HeuristicLab.Parameters;
|
|---|
| 29 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
|---|
| 30 |
|
|---|
| 31 | namespace HeuristicLab.Problems.DataAnalysis.Symbolic.Regression {
|
|---|
| 32 | [Item("Constant Optimization Evaluator", "Calculates Pearson R² of a symbolic regression solution and optimizes the constant used.")]
|
|---|
| 33 | [StorableClass]
|
|---|
| 34 | public class SymbolicRegressionConstantOptimizationEvaluator : SymbolicRegressionSingleObjectiveEvaluator {
|
|---|
| 35 | private const string ConstantOptimizationIterationsParameterName = "ConstantOptimizationIterations";
|
|---|
| 36 | private const string ConstantOptimizationImprovementParameterName = "ConstantOptimizationImprovement";
|
|---|
| 37 | private const string ConstantOptimizationProbabilityParameterName = "ConstantOptimizationProbability";
|
|---|
| 38 | private const string ConstantOptimizationRowsPercentageParameterName = "ConstantOptimizationRowsPercentage";
|
|---|
| 39 |
|
|---|
| 40 | private const string EvaluatedTreesResultName = "EvaluatedTrees";
|
|---|
| 41 | private const string EvaluatedTreeNodesResultName = "EvaluatedTreeNodes";
|
|---|
| 42 |
|
|---|
| 43 | public ILookupParameter<IntValue> EvaluatedTreesParameter {
|
|---|
| 44 | get { return (ILookupParameter<IntValue>)Parameters[EvaluatedTreesResultName]; }
|
|---|
| 45 | }
|
|---|
| 46 | public ILookupParameter<IntValue> EvaluatedTreeNodesParameter {
|
|---|
| 47 | get { return (ILookupParameter<IntValue>)Parameters[EvaluatedTreeNodesResultName]; }
|
|---|
| 48 | }
|
|---|
| 49 |
|
|---|
| 50 | public IFixedValueParameter<IntValue> ConstantOptimizationIterationsParameter {
|
|---|
| 51 | get { return (IFixedValueParameter<IntValue>)Parameters[ConstantOptimizationIterationsParameterName]; }
|
|---|
| 52 | }
|
|---|
| 53 | public IFixedValueParameter<DoubleValue> ConstantOptimizationImprovementParameter {
|
|---|
| 54 | get { return (IFixedValueParameter<DoubleValue>)Parameters[ConstantOptimizationImprovementParameterName]; }
|
|---|
| 55 | }
|
|---|
| 56 | public IFixedValueParameter<PercentValue> ConstantOptimizationProbabilityParameter {
|
|---|
| 57 | get { return (IFixedValueParameter<PercentValue>)Parameters[ConstantOptimizationProbabilityParameterName]; }
|
|---|
| 58 | }
|
|---|
| 59 | public IFixedValueParameter<PercentValue> ConstantOptimizationRowsPercentageParameter {
|
|---|
| 60 | get { return (IFixedValueParameter<PercentValue>)Parameters[ConstantOptimizationRowsPercentageParameterName]; }
|
|---|
| 61 | }
|
|---|
| 62 |
|
|---|
| 63 | public IntValue ConstantOptimizationIterations {
|
|---|
| 64 | get { return ConstantOptimizationIterationsParameter.Value; }
|
|---|
| 65 | }
|
|---|
| 66 | public DoubleValue ConstantOptimizationImprovement {
|
|---|
| 67 | get { return ConstantOptimizationImprovementParameter.Value; }
|
|---|
| 68 | }
|
|---|
| 69 | public PercentValue ConstantOptimizationProbability {
|
|---|
| 70 | get { return ConstantOptimizationProbabilityParameter.Value; }
|
|---|
| 71 | }
|
|---|
| 72 | public PercentValue ConstantOptimizationRowsPercentage {
|
|---|
| 73 | get { return ConstantOptimizationRowsPercentageParameter.Value; }
|
|---|
| 74 | }
|
|---|
| 75 |
|
|---|
| 76 | public override bool Maximization {
|
|---|
| 77 | get { return true; }
|
|---|
| 78 | }
|
|---|
| 79 |
|
|---|
| 80 | [StorableConstructor]
|
|---|
| 81 | protected SymbolicRegressionConstantOptimizationEvaluator(bool deserializing) : base(deserializing) { }
|
|---|
| 82 | protected SymbolicRegressionConstantOptimizationEvaluator(SymbolicRegressionConstantOptimizationEvaluator original, Cloner cloner)
|
|---|
| 83 | : base(original, cloner) {
|
|---|
| 84 | }
|
|---|
| 85 | public SymbolicRegressionConstantOptimizationEvaluator()
|
|---|
| 86 | : base() {
|
|---|
| 87 | Parameters.Add(new FixedValueParameter<IntValue>(ConstantOptimizationIterationsParameterName, "Determines how many iterations should be calculated while optimizing the constant of a symbolic expression tree (0 indicates other or default stopping criterion).", new IntValue(3), true));
|
|---|
| 88 | Parameters.Add(new FixedValueParameter<DoubleValue>(ConstantOptimizationImprovementParameterName, "Determines the relative improvement which must be achieved in the constant optimization to continue with it (0 indicates other or default stopping criterion).", new DoubleValue(0), true));
|
|---|
| 89 | Parameters.Add(new FixedValueParameter<PercentValue>(ConstantOptimizationProbabilityParameterName, "Determines the probability that the constants are optimized", new PercentValue(1), true));
|
|---|
| 90 | Parameters.Add(new FixedValueParameter<PercentValue>(ConstantOptimizationRowsPercentageParameterName, "Determines the percentage of the rows which should be used for constant optimization", new PercentValue(1), true));
|
|---|
| 91 |
|
|---|
| 92 | Parameters.Add(new LookupParameter<IntValue>(EvaluatedTreesResultName));
|
|---|
| 93 | Parameters.Add(new LookupParameter<IntValue>(EvaluatedTreeNodesResultName));
|
|---|
| 94 | }
|
|---|
| 95 |
|
|---|
| 96 | public override IDeepCloneable Clone(Cloner cloner) {
|
|---|
| 97 | return new SymbolicRegressionConstantOptimizationEvaluator(this, cloner);
|
|---|
| 98 | }
|
|---|
| 99 |
|
|---|
| 100 | public override IOperation Apply() {
|
|---|
| 101 | AddResults();
|
|---|
| 102 | int seed = RandomParameter.ActualValue.Next();
|
|---|
| 103 | var solution = SymbolicExpressionTreeParameter.ActualValue;
|
|---|
| 104 | double quality;
|
|---|
| 105 | if (RandomParameter.ActualValue.NextDouble() < ConstantOptimizationProbability.Value) {
|
|---|
| 106 | IEnumerable<int> constantOptimizationRows = GenerateRowsToEvaluate(ConstantOptimizationRowsPercentage.Value);
|
|---|
| 107 | quality = OptimizeConstants(SymbolicDataAnalysisTreeInterpreterParameter.ActualValue, solution, ProblemDataParameter.ActualValue,
|
|---|
| 108 | constantOptimizationRows, ConstantOptimizationImprovement.Value, ConstantOptimizationIterations.Value, 0.001,
|
|---|
| 109 | EstimationLimitsParameter.ActualValue.Upper, EstimationLimitsParameter.ActualValue.Lower,
|
|---|
| 110 | EvaluatedTreesParameter.ActualValue, EvaluatedTreeNodesParameter.ActualValue);
|
|---|
| 111 | if (ConstantOptimizationRowsPercentage.Value != RelativeNumberOfEvaluatedSamplesParameter.ActualValue.Value) {
|
|---|
| 112 | var evaluationRows = GenerateRowsToEvaluate();
|
|---|
| 113 | quality = SymbolicRegressionSingleObjectivePearsonRSquaredEvaluator.Calculate(SymbolicDataAnalysisTreeInterpreterParameter.ActualValue, solution, EstimationLimitsParameter.ActualValue.Lower, EstimationLimitsParameter.ActualValue.Upper, ProblemDataParameter.ActualValue, evaluationRows);
|
|---|
| 114 | }
|
|---|
| 115 | } else {
|
|---|
| 116 | var evaluationRows = GenerateRowsToEvaluate();
|
|---|
| 117 | quality = SymbolicRegressionSingleObjectivePearsonRSquaredEvaluator.Calculate(SymbolicDataAnalysisTreeInterpreterParameter.ActualValue, solution, EstimationLimitsParameter.ActualValue.Lower, EstimationLimitsParameter.ActualValue.Upper, ProblemDataParameter.ActualValue, evaluationRows);
|
|---|
| 118 | }
|
|---|
| 119 | QualityParameter.ActualValue = new DoubleValue(quality);
|
|---|
| 120 | EvaluatedTreesParameter.ActualValue.Value += 1;
|
|---|
| 121 | EvaluatedTreeNodesParameter.ActualValue.Value += solution.Length;
|
|---|
| 122 |
|
|---|
| 123 | if (Successor != null)
|
|---|
| 124 | return ExecutionContext.CreateOperation(Successor);
|
|---|
| 125 | else
|
|---|
| 126 | return null;
|
|---|
| 127 | }
|
|---|
| 128 |
|
|---|
| 129 | private void AddResults() {
|
|---|
| 130 | if (EvaluatedTreesParameter.ActualValue == null) {
|
|---|
| 131 | var scope = ExecutionContext.Scope;
|
|---|
| 132 | while (scope.Parent != null)
|
|---|
| 133 | scope = scope.Parent;
|
|---|
| 134 | scope.Variables.Add(new Core.Variable(EvaluatedTreesResultName, new IntValue()));
|
|---|
| 135 | }
|
|---|
| 136 | if (EvaluatedTreeNodesParameter.ActualValue == null) {
|
|---|
| 137 | var scope = ExecutionContext.Scope;
|
|---|
| 138 | while (scope.Parent != null)
|
|---|
| 139 | scope = scope.Parent;
|
|---|
| 140 | scope.Variables.Add(new Core.Variable(EvaluatedTreeNodesResultName, new IntValue()));
|
|---|
| 141 | }
|
|---|
| 142 | }
|
|---|
| 143 |
|
|---|
| 144 | public override double Evaluate(IExecutionContext context, ISymbolicExpressionTree tree, IRegressionProblemData problemData, IEnumerable<int> rows) {
|
|---|
| 145 | SymbolicDataAnalysisTreeInterpreterParameter.ExecutionContext = context;
|
|---|
| 146 | EstimationLimitsParameter.ExecutionContext = context;
|
|---|
| 147 |
|
|---|
| 148 | double r2 = SymbolicRegressionSingleObjectivePearsonRSquaredEvaluator.Calculate(SymbolicDataAnalysisTreeInterpreterParameter.ActualValue, tree, EstimationLimitsParameter.ActualValue.Lower, EstimationLimitsParameter.ActualValue.Upper, problemData, rows);
|
|---|
| 149 |
|
|---|
| 150 | SymbolicDataAnalysisTreeInterpreterParameter.ExecutionContext = null;
|
|---|
| 151 | EstimationLimitsParameter.ExecutionContext = null;
|
|---|
| 152 |
|
|---|
| 153 | return r2;
|
|---|
| 154 | }
|
|---|
| 155 |
|
|---|
| 156 | public static double OptimizeConstants(ISymbolicDataAnalysisExpressionTreeInterpreter interpreter, ISymbolicExpressionTree tree, IRegressionProblemData problemData,
|
|---|
| 157 | IEnumerable<int> rows, double improvement, int iterations, double differentialStep, double upperEstimationLimit = double.MaxValue, double lowerEstimationLimit = double.MinValue, IntValue evaluatedTrees = null, IntValue evaluatedTreeNodes = null) {
|
|---|
| 158 | List<SymbolicExpressionTreeTerminalNode> terminalNodes = tree.Root.IterateNodesPrefix().OfType<SymbolicExpressionTreeTerminalNode>().ToList();
|
|---|
| 159 | double[] c = new double[terminalNodes.Count];
|
|---|
| 160 | int treeLength = tree.Length;
|
|---|
| 161 |
|
|---|
| 162 | //extract inital constants
|
|---|
| 163 | for (int i = 0; i < terminalNodes.Count; i++) {
|
|---|
| 164 | ConstantTreeNode constantTreeNode = terminalNodes[i] as ConstantTreeNode;
|
|---|
| 165 | if (constantTreeNode != null) c[i] = constantTreeNode.Value;
|
|---|
| 166 | VariableTreeNode variableTreeNode = terminalNodes[i] as VariableTreeNode;
|
|---|
| 167 | if (variableTreeNode != null) c[i] = variableTreeNode.Weight;
|
|---|
| 168 | }
|
|---|
| 169 |
|
|---|
| 170 | double epsg = 0;
|
|---|
| 171 | double epsf = improvement;
|
|---|
| 172 | double epsx = 0;
|
|---|
| 173 | int maxits = iterations;
|
|---|
| 174 | double diffstep = differentialStep;
|
|---|
| 175 |
|
|---|
| 176 | alglib.minlmstate state;
|
|---|
| 177 | alglib.minlmreport report;
|
|---|
| 178 |
|
|---|
| 179 | alglib.minlmcreatev(1, c, diffstep, out state);
|
|---|
| 180 | alglib.minlmsetcond(state, epsg, epsf, epsx, maxits);
|
|---|
| 181 | alglib.minlmoptimize(state, CreateCallBack(interpreter, tree, problemData, rows, upperEstimationLimit, lowerEstimationLimit, treeLength, evaluatedTrees, evaluatedTreeNodes), null, terminalNodes);
|
|---|
| 182 | alglib.minlmresults(state, out c, out report);
|
|---|
| 183 |
|
|---|
| 184 | for (int i = 0; i < c.Length; i++) {
|
|---|
| 185 | ConstantTreeNode constantTreeNode = terminalNodes[i] as ConstantTreeNode;
|
|---|
| 186 | if (constantTreeNode != null) constantTreeNode.Value = c[i];
|
|---|
| 187 | VariableTreeNode variableTreeNode = terminalNodes[i] as VariableTreeNode;
|
|---|
| 188 | if (variableTreeNode != null) variableTreeNode.Weight = c[i];
|
|---|
| 189 | }
|
|---|
| 190 |
|
|---|
| 191 | return (state.fi[0] - 1) * -1;
|
|---|
| 192 | }
|
|---|
| 193 |
|
|---|
| 194 | private static alglib.ndimensional_fvec CreateCallBack(ISymbolicDataAnalysisExpressionTreeInterpreter interpreter, ISymbolicExpressionTree tree, IRegressionProblemData problemData, IEnumerable<int> rows, double upperEstimationLimit, double lowerEstimationLimit, int treeLength, IntValue evaluatedTrees = null, IntValue evaluatedTreeNodes = null) {
|
|---|
| 195 | return (double[] arg, double[] fi, object obj) => {
|
|---|
| 196 | // update constants of tree
|
|---|
| 197 | List<SymbolicExpressionTreeTerminalNode> terminalNodes = (List<SymbolicExpressionTreeTerminalNode>)obj;
|
|---|
| 198 | for (int i = 0; i < terminalNodes.Count; i++) {
|
|---|
| 199 | ConstantTreeNode constantTreeNode = terminalNodes[i] as ConstantTreeNode;
|
|---|
| 200 | if (constantTreeNode != null) constantTreeNode.Value = arg[i];
|
|---|
| 201 | VariableTreeNode variableTreeNode = terminalNodes[i] as VariableTreeNode;
|
|---|
| 202 | if (variableTreeNode != null) variableTreeNode.Weight = arg[i];
|
|---|
| 203 | }
|
|---|
| 204 |
|
|---|
| 205 | double quality = SymbolicRegressionSingleObjectivePearsonRSquaredEvaluator.Calculate(interpreter, tree, lowerEstimationLimit, upperEstimationLimit, problemData, rows);
|
|---|
| 206 |
|
|---|
| 207 | fi[0] = 1 - quality;
|
|---|
| 208 | if (evaluatedTrees != null) evaluatedTrees.Value++;
|
|---|
| 209 | if (evaluatedTreeNodes != null) evaluatedTreeNodes.Value += treeLength;
|
|---|
| 210 | };
|
|---|
| 211 | }
|
|---|
| 212 |
|
|---|
| 213 | }
|
|---|
| 214 | }
|
|---|
