[14024] | 1 | #region License Information
|
---|
| 2 | /* HeuristicLab
|
---|
| 3 | * Copyright (C) 2002-2016 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;
|
---|
| 23 | using System.Collections.Generic;
|
---|
| 24 | using System.Linq;
|
---|
| 25 | using HeuristicLab.Common;
|
---|
| 26 | using HeuristicLab.Core;
|
---|
| 27 | using HeuristicLab.Data;
|
---|
| 28 | using HeuristicLab.Parameters;
|
---|
| 29 | using HeuristicLab.Optimization;
|
---|
| 30 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
---|
| 31 | using HeuristicLab.Problems.DataAnalysis;
|
---|
| 32 | using HeuristicLab.Problems.DataAnalysis.Symbolic;
|
---|
| 33 | using HeuristicLab.Problems.DataAnalysis.Symbolic.Regression;
|
---|
[14277] | 34 | using HeuristicLab.Random;
|
---|
[14024] | 35 |
|
---|
| 36 | namespace HeuristicLab.Algorithms.DataAnalysis {
|
---|
| 37 | /// <summary>
|
---|
| 38 | /// Nonlinear regression data analysis algorithm.
|
---|
| 39 | /// </summary>
|
---|
| 40 | [Item("Nonlinear Regression (NLR)", "Nonlinear regression (curve fitting) data analysis algorithm (wrapper for ALGLIB).")]
|
---|
| 41 | [Creatable(CreatableAttribute.Categories.DataAnalysisRegression, Priority = 120)]
|
---|
| 42 | [StorableClass]
|
---|
| 43 | public sealed class NonlinearRegression : FixedDataAnalysisAlgorithm<IRegressionProblem> {
|
---|
[14109] | 44 | private const string RegressionSolutionResultName = "Regression solution";
|
---|
[14024] | 45 | private const string ModelStructureParameterName = "Model structure";
|
---|
| 46 | private const string IterationsParameterName = "Iterations";
|
---|
[14277] | 47 | private const string RestartsParameterName = "Restarts";
|
---|
| 48 | private const string SetSeedRandomlyParameterName = "SetSeedRandomly";
|
---|
| 49 | private const string SeedParameterName = "Seed";
|
---|
[14024] | 50 |
|
---|
| 51 | public IFixedValueParameter<StringValue> ModelStructureParameter {
|
---|
| 52 | get { return (IFixedValueParameter<StringValue>)Parameters[ModelStructureParameterName]; }
|
---|
| 53 | }
|
---|
| 54 | public IFixedValueParameter<IntValue> IterationsParameter {
|
---|
| 55 | get { return (IFixedValueParameter<IntValue>)Parameters[IterationsParameterName]; }
|
---|
| 56 | }
|
---|
| 57 |
|
---|
[14277] | 58 | public IFixedValueParameter<BoolValue> SetSeedRandomlyParameter {
|
---|
| 59 | get { return (IFixedValueParameter<BoolValue>)Parameters[SetSeedRandomlyParameterName]; }
|
---|
| 60 | }
|
---|
| 61 |
|
---|
| 62 | public IFixedValueParameter<IntValue> SeedParameter {
|
---|
| 63 | get { return (IFixedValueParameter<IntValue>)Parameters[SeedParameterName]; }
|
---|
| 64 | }
|
---|
| 65 |
|
---|
| 66 | public IFixedValueParameter<IntValue> RestartsParameter {
|
---|
| 67 | get { return (IFixedValueParameter<IntValue>)Parameters[RestartsParameterName]; }
|
---|
| 68 | }
|
---|
| 69 |
|
---|
[14024] | 70 | public string ModelStructure {
|
---|
| 71 | get { return ModelStructureParameter.Value.Value; }
|
---|
| 72 | set { ModelStructureParameter.Value.Value = value; }
|
---|
| 73 | }
|
---|
| 74 |
|
---|
| 75 | public int Iterations {
|
---|
| 76 | get { return IterationsParameter.Value.Value; }
|
---|
| 77 | set { IterationsParameter.Value.Value = value; }
|
---|
| 78 | }
|
---|
| 79 |
|
---|
[14277] | 80 | public int Restarts {
|
---|
| 81 | get { return RestartsParameter.Value.Value; }
|
---|
| 82 | set { RestartsParameter.Value.Value = value; }
|
---|
| 83 | }
|
---|
[14024] | 84 |
|
---|
[14277] | 85 | public int Seed {
|
---|
| 86 | get { return SeedParameter.Value.Value; }
|
---|
| 87 | set { SeedParameter.Value.Value = value; }
|
---|
| 88 | }
|
---|
| 89 |
|
---|
| 90 | public bool SetSeedRandomly {
|
---|
| 91 | get { return SetSeedRandomlyParameter.Value.Value; }
|
---|
| 92 | set { SetSeedRandomlyParameter.Value.Value = value; }
|
---|
| 93 | }
|
---|
| 94 |
|
---|
[14024] | 95 | [StorableConstructor]
|
---|
| 96 | private NonlinearRegression(bool deserializing) : base(deserializing) { }
|
---|
| 97 | private NonlinearRegression(NonlinearRegression original, Cloner cloner)
|
---|
| 98 | : base(original, cloner) {
|
---|
| 99 | }
|
---|
| 100 | public NonlinearRegression()
|
---|
| 101 | : base() {
|
---|
| 102 | Problem = new RegressionProblem();
|
---|
| 103 | Parameters.Add(new FixedValueParameter<StringValue>(ModelStructureParameterName, "The function for which the parameters must be fit (only numeric constants are tuned).", new StringValue("1.0 * x*x + 0.0")));
|
---|
| 104 | Parameters.Add(new FixedValueParameter<IntValue>(IterationsParameterName, "The maximum number of iterations for constants optimization.", new IntValue(200)));
|
---|
[14277] | 105 | Parameters.Add(new FixedValueParameter<IntValue>(RestartsParameterName, "The number of independent random restarts", new IntValue(10)));
|
---|
| 106 | Parameters.Add(new FixedValueParameter<IntValue>(SeedParameterName, "The PRNG seed value.", new IntValue()));
|
---|
| 107 | Parameters.Add(new FixedValueParameter<BoolValue>(SetSeedRandomlyParameterName, "Switch to determine if the random number seed should be initialized randomly.", new BoolValue(true)));
|
---|
[14024] | 108 | }
|
---|
[14277] | 109 |
|
---|
[14024] | 110 | [StorableHook(HookType.AfterDeserialization)]
|
---|
[14277] | 111 | private void AfterDeserialization() {
|
---|
| 112 | // BackwardsCompatibility3.3
|
---|
| 113 | #region Backwards compatible code, remove with 3.4
|
---|
| 114 | if (!Parameters.ContainsKey(RestartsParameterName))
|
---|
| 115 | Parameters.Add(new FixedValueParameter<IntValue>(RestartsParameterName, "The number of independent random restarts", new IntValue(1)));
|
---|
| 116 | if (!Parameters.ContainsKey(SeedParameterName))
|
---|
| 117 | Parameters.Add(new FixedValueParameter<IntValue>(SeedParameterName, "The PRNG seed value.", new IntValue()));
|
---|
| 118 | if (!Parameters.ContainsKey(SetSeedRandomlyParameterName))
|
---|
| 119 | Parameters.Add(new FixedValueParameter<BoolValue>(SetSeedRandomlyParameterName, "Switch to determine if the random number seed should be initialized randomly.", new BoolValue(true)));
|
---|
| 120 | #endregion
|
---|
| 121 | }
|
---|
[14024] | 122 |
|
---|
| 123 | public override IDeepCloneable Clone(Cloner cloner) {
|
---|
| 124 | return new NonlinearRegression(this, cloner);
|
---|
| 125 | }
|
---|
| 126 |
|
---|
| 127 | #region nonlinear regression
|
---|
| 128 | protected override void Run() {
|
---|
[14277] | 129 | if (SetSeedRandomly) Seed = (new System.Random()).Next();
|
---|
| 130 | var rand = new MersenneTwister((uint)Seed);
|
---|
| 131 | IRegressionSolution bestSolution = null;
|
---|
| 132 | for (int r = 0; r < Restarts; r++) {
|
---|
| 133 | var solution = CreateRegressionSolution(Problem.ProblemData, ModelStructure, Iterations, rand);
|
---|
| 134 | if (bestSolution == null || solution.TrainingRootMeanSquaredError < bestSolution.TrainingRootMeanSquaredError) {
|
---|
| 135 | bestSolution = solution;
|
---|
| 136 | }
|
---|
| 137 | }
|
---|
| 138 |
|
---|
| 139 | Results.Add(new Result(RegressionSolutionResultName, "The nonlinear regression solution.", bestSolution));
|
---|
| 140 | Results.Add(new Result("Root mean square error (train)", "The root of the mean of squared errors of the regression solution on the training set.", new DoubleValue(bestSolution.TrainingRootMeanSquaredError)));
|
---|
| 141 | Results.Add(new Result("Root mean square error (test)", "The root of the mean of squared errors of the regression solution on the test set.", new DoubleValue(bestSolution.TestRootMeanSquaredError)));
|
---|
| 142 |
|
---|
[14024] | 143 | }
|
---|
| 144 |
|
---|
[14277] | 145 | /// <summary>
|
---|
| 146 | /// Fits a model to the data by optimizing the numeric constants.
|
---|
| 147 | /// Model is specified as infix expression containing variable names and numbers.
|
---|
| 148 | /// The starting point for the numeric constants is initialized randomly if a random number generator is specified (~N(0,1)). Otherwise the user specified constants are
|
---|
| 149 | /// used as a starting point.
|
---|
| 150 | /// </summary>
|
---|
| 151 | /// <param name="problemData">Training and test data</param>
|
---|
| 152 | /// <param name="modelStructure">The function as infix expression</param>
|
---|
| 153 | /// <param name="maxIterations">Number of constant optimization iterations (using Levenberg-Marquardt algorithm)</param>
|
---|
| 154 | /// <param name="random">Optional random number generator for random initialization of numeric constants.</param>
|
---|
| 155 | /// <returns></returns>
|
---|
| 156 | public static ISymbolicRegressionSolution CreateRegressionSolution(IRegressionProblemData problemData, string modelStructure, int maxIterations, IRandom random = null) {
|
---|
[14024] | 157 | var parser = new InfixExpressionParser();
|
---|
| 158 | var tree = parser.Parse(modelStructure);
|
---|
[14251] | 159 | // parser handles double and string variables equally by creating a VariableTreeNode
|
---|
| 160 | // post-process to replace VariableTreeNodes by FactorVariableTreeNodes for all string variables
|
---|
| 161 | var factorSymbol = new FactorVariable();
|
---|
| 162 | factorSymbol.VariableNames =
|
---|
| 163 | problemData.AllowedInputVariables.Where(name => problemData.Dataset.VariableHasType<string>(name));
|
---|
| 164 | factorSymbol.AllVariableNames = factorSymbol.VariableNames;
|
---|
| 165 | factorSymbol.VariableValues =
|
---|
| 166 | factorSymbol.VariableNames.Select(name => new KeyValuePair<string, List<string>>(name, problemData.Dataset.GetReadOnlyStringValues(name).Distinct().ToList()));
|
---|
| 167 |
|
---|
| 168 | foreach (var parent in tree.IterateNodesPrefix().ToArray()) {
|
---|
| 169 | for (int i = 0; i < parent.SubtreeCount; i++) {
|
---|
| 170 | var child = parent.GetSubtree(i) as VariableTreeNode;
|
---|
| 171 | if (child != null && factorSymbol.VariableNames.Contains(child.VariableName)) {
|
---|
| 172 | parent.RemoveSubtree(i);
|
---|
| 173 | var factorTreeNode = (FactorVariableTreeNode)factorSymbol.CreateTreeNode();
|
---|
| 174 | factorTreeNode.VariableName = child.VariableName;
|
---|
| 175 | factorTreeNode.Weights =
|
---|
| 176 | factorTreeNode.Symbol.GetVariableValues(factorTreeNode.VariableName).Select(_ => 1.0).ToArray(); // weight = 1.0 for each value
|
---|
| 177 | parent.InsertSubtree(i, factorTreeNode);
|
---|
| 178 | }
|
---|
| 179 | }
|
---|
| 180 | }
|
---|
| 181 |
|
---|
[14024] | 182 | if (!SymbolicRegressionConstantOptimizationEvaluator.CanOptimizeConstants(tree)) throw new ArgumentException("The optimizer does not support the specified model structure.");
|
---|
| 183 |
|
---|
[14277] | 184 | // initialize constants randomly
|
---|
| 185 | if (random != null) {
|
---|
| 186 | foreach (var node in tree.IterateNodesPrefix().OfType<ConstantTreeNode>()) {
|
---|
| 187 | node.Value = NormalDistributedRandom.NextDouble(random, 0, 1);
|
---|
| 188 | }
|
---|
| 189 | }
|
---|
[14024] | 190 | var interpreter = new SymbolicDataAnalysisExpressionTreeLinearInterpreter();
|
---|
[14277] | 191 |
|
---|
[14251] | 192 | SymbolicRegressionConstantOptimizationEvaluator.OptimizeConstants(interpreter, tree, problemData, problemData.TrainingIndices,
|
---|
[14024] | 193 | applyLinearScaling: false, maxIterations: maxIterations,
|
---|
| 194 | updateVariableWeights: false, updateConstantsInTree: true);
|
---|
| 195 |
|
---|
| 196 | var scaledModel = new SymbolicRegressionModel(problemData.TargetVariable, tree, (ISymbolicDataAnalysisExpressionTreeInterpreter)interpreter.Clone());
|
---|
| 197 | scaledModel.Scale(problemData);
|
---|
| 198 | SymbolicRegressionSolution solution = new SymbolicRegressionSolution(scaledModel, (IRegressionProblemData)problemData.Clone());
|
---|
| 199 | solution.Model.Name = "Regression Model";
|
---|
| 200 | solution.Name = "Regression Solution";
|
---|
| 201 | return solution;
|
---|
| 202 | }
|
---|
| 203 | #endregion
|
---|
| 204 | }
|
---|
| 205 | }
|
---|