1 | #region License Information
|
---|
2 | /* HeuristicLab
|
---|
3 | * Copyright (C) 2002-2019 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 HEAL.Attic;
|
---|
26 | using HeuristicLab.Common;
|
---|
27 | using HeuristicLab.Core;
|
---|
28 | using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
|
---|
29 | using HeuristicLab.Parameters;
|
---|
30 | using HeuristicLab.Problems.DataAnalysis;
|
---|
31 | using HeuristicLab.Problems.Instances;
|
---|
32 |
|
---|
33 |
|
---|
34 | namespace HeuristicLab.Problems.GeneticProgramming.BasicSymbolicRegression {
|
---|
35 | [Item("Koza-style Symbolic Regression", "An implementation of symbolic regression without bells-and-whistles. Use \"Symbolic Regression Problem (single-objective)\" if you want to use all features.")]
|
---|
36 | [Creatable(CreatableAttribute.Categories.GeneticProgrammingProblems, Priority = 900)]
|
---|
37 | [StorableType("72011B73-28C6-4D5E-BEDF-27425BC87B9C")]
|
---|
38 | public sealed class Problem : SymbolicExpressionTreeProblem, IRegressionProblem, IProblemInstanceConsumer<IRegressionProblemData>, IProblemInstanceExporter<IRegressionProblemData> {
|
---|
39 |
|
---|
40 | #region parameter names
|
---|
41 | private const string ProblemDataParameterName = "ProblemData";
|
---|
42 | #endregion
|
---|
43 |
|
---|
44 | #region Parameter Properties
|
---|
45 | IParameter IDataAnalysisProblem.ProblemDataParameter { get { return ProblemDataParameter; } }
|
---|
46 |
|
---|
47 | public IValueParameter<IRegressionProblemData> ProblemDataParameter {
|
---|
48 | get { return (IValueParameter<IRegressionProblemData>)Parameters[ProblemDataParameterName]; }
|
---|
49 | }
|
---|
50 | #endregion
|
---|
51 |
|
---|
52 | #region Properties
|
---|
53 | public IRegressionProblemData ProblemData {
|
---|
54 | get { return ProblemDataParameter.Value; }
|
---|
55 | set { ProblemDataParameter.Value = value; }
|
---|
56 | }
|
---|
57 | IDataAnalysisProblemData IDataAnalysisProblem.ProblemData { get { return ProblemData; } }
|
---|
58 | #endregion
|
---|
59 |
|
---|
60 | public event EventHandler ProblemDataChanged;
|
---|
61 |
|
---|
62 | public override bool Maximization {
|
---|
63 | get { return true; }
|
---|
64 | }
|
---|
65 |
|
---|
66 | #region item cloning and persistence
|
---|
67 | // persistence
|
---|
68 | [StorableConstructor]
|
---|
69 | private Problem(StorableConstructorFlag _) : base(_) { }
|
---|
70 | [StorableHook(HookType.AfterDeserialization)]
|
---|
71 | private void AfterDeserialization() {
|
---|
72 | RegisterEventHandlers();
|
---|
73 | }
|
---|
74 |
|
---|
75 | // cloning
|
---|
76 | private Problem(Problem original, Cloner cloner)
|
---|
77 | : base(original, cloner) {
|
---|
78 | RegisterEventHandlers();
|
---|
79 | }
|
---|
80 | public override IDeepCloneable Clone(Cloner cloner) { return new Problem(this, cloner); }
|
---|
81 | #endregion
|
---|
82 |
|
---|
83 | public Problem() : base(new SymbolicExpressionTreeEncoding()) {
|
---|
84 | Parameters.Add(new ValueParameter<IRegressionProblemData>(ProblemDataParameterName, "The data for the regression problem", new RegressionProblemData()));
|
---|
85 |
|
---|
86 | Encoding.TreeLength = 100;
|
---|
87 | Encoding.TreeDepth = 17;
|
---|
88 |
|
---|
89 | UpdateGrammar();
|
---|
90 | RegisterEventHandlers();
|
---|
91 | }
|
---|
92 |
|
---|
93 |
|
---|
94 | public override double Evaluate(ISymbolicExpressionTree tree, IRandom random) {
|
---|
95 | // Doesn't use classes from HeuristicLab.Problems.DataAnalysis.Symbolic to make sure that the implementation can be fully understood easily.
|
---|
96 | // HeuristicLab.Problems.DataAnalysis.Symbolic would already provide all the necessary functionality (esp. interpreter) but at a much higher complexity.
|
---|
97 | // Another argument is that we don't need a reference to HeuristicLab.Problems.DataAnalysis.Symbolic
|
---|
98 |
|
---|
99 | var problemData = ProblemData;
|
---|
100 | var rows = ProblemData.TrainingIndices.ToArray();
|
---|
101 | var target = problemData.Dataset.GetDoubleValues(problemData.TargetVariable, rows);
|
---|
102 | var predicted = Interpret(tree, problemData.Dataset, rows);
|
---|
103 |
|
---|
104 | OnlineCalculatorError errorState;
|
---|
105 | var r = OnlinePearsonsRCalculator.Calculate(target, predicted, out errorState);
|
---|
106 | if (errorState != OnlineCalculatorError.None) r = 0;
|
---|
107 | return r * r;
|
---|
108 | }
|
---|
109 |
|
---|
110 | private IEnumerable<double> Interpret(ISymbolicExpressionTree tree, IDataset dataset, IEnumerable<int> rows) {
|
---|
111 | // skip programRoot and startSymbol
|
---|
112 | return InterpretRec(tree.Root.GetSubtree(0).GetSubtree(0), dataset, rows);
|
---|
113 | }
|
---|
114 |
|
---|
115 | private IEnumerable<double> InterpretRec(ISymbolicExpressionTreeNode node, IDataset dataset, IEnumerable<int> rows) {
|
---|
116 | Func<ISymbolicExpressionTreeNode, ISymbolicExpressionTreeNode, Func<double, double, double>, IEnumerable<double>> binaryEval =
|
---|
117 | (left, right, f) => InterpretRec(left, dataset, rows).Zip(InterpretRec(right, dataset, rows), f);
|
---|
118 |
|
---|
119 | switch (node.Symbol.Name) {
|
---|
120 | case "+": return binaryEval(node.GetSubtree(0), node.GetSubtree(1), (x, y) => x + y);
|
---|
121 | case "*": return binaryEval(node.GetSubtree(0), node.GetSubtree(1), (x, y) => x * y);
|
---|
122 | case "-": return binaryEval(node.GetSubtree(0), node.GetSubtree(1), (x, y) => x - y);
|
---|
123 | case "%": return binaryEval(node.GetSubtree(0), node.GetSubtree(1), (x, y) => y.IsAlmost(0.0) ? 0.0 : x / y); // protected division
|
---|
124 | default: {
|
---|
125 | double erc;
|
---|
126 | if (double.TryParse(node.Symbol.Name, out erc)) {
|
---|
127 | return rows.Select(_ => erc);
|
---|
128 | } else {
|
---|
129 | // assume that this is a variable name
|
---|
130 | return dataset.GetDoubleValues(node.Symbol.Name, rows);
|
---|
131 | }
|
---|
132 | }
|
---|
133 | }
|
---|
134 | }
|
---|
135 |
|
---|
136 |
|
---|
137 | #region events
|
---|
138 | private void RegisterEventHandlers() {
|
---|
139 | ProblemDataParameter.ValueChanged += new EventHandler(ProblemDataParameter_ValueChanged);
|
---|
140 | if (ProblemDataParameter.Value != null) ProblemDataParameter.Value.Changed += new EventHandler(ProblemData_Changed);
|
---|
141 | }
|
---|
142 |
|
---|
143 | private void ProblemDataParameter_ValueChanged(object sender, EventArgs e) {
|
---|
144 | ProblemDataParameter.Value.Changed += new EventHandler(ProblemData_Changed);
|
---|
145 | OnProblemDataChanged();
|
---|
146 | OnReset();
|
---|
147 | }
|
---|
148 |
|
---|
149 | private void ProblemData_Changed(object sender, EventArgs e) {
|
---|
150 | OnReset();
|
---|
151 | }
|
---|
152 |
|
---|
153 | private void OnProblemDataChanged() {
|
---|
154 | UpdateGrammar();
|
---|
155 |
|
---|
156 | var handler = ProblemDataChanged;
|
---|
157 | if (handler != null) handler(this, EventArgs.Empty);
|
---|
158 | }
|
---|
159 |
|
---|
160 | private void UpdateGrammar() {
|
---|
161 | // whenever ProblemData is changed we create a new grammar with the necessary symbols
|
---|
162 | var g = new SimpleSymbolicExpressionGrammar();
|
---|
163 | g.AddSymbols(new[] { "+", "*", "%", "-" }, 2, 2); // % is protected division 1/0 := 0
|
---|
164 |
|
---|
165 | foreach (var variableName in ProblemData.AllowedInputVariables)
|
---|
166 | g.AddTerminalSymbol(variableName);
|
---|
167 |
|
---|
168 | // generate ephemeral random consts in the range [-10..+10[ (2*number of variables)
|
---|
169 | var rand = new System.Random();
|
---|
170 | for (int i = 0; i < ProblemData.AllowedInputVariables.Count() * 2; i++) {
|
---|
171 | string newErcSy;
|
---|
172 | do {
|
---|
173 | newErcSy = string.Format("{0:F2}", rand.NextDouble() * 20 - 10);
|
---|
174 | } while (g.Symbols.Any(sy => sy.Name == newErcSy)); // it might happen that we generate the same constant twice
|
---|
175 | g.AddTerminalSymbol(newErcSy);
|
---|
176 | }
|
---|
177 |
|
---|
178 | Encoding.Grammar = g;
|
---|
179 | }
|
---|
180 | #endregion
|
---|
181 |
|
---|
182 | #region Import & Export
|
---|
183 | public void Load(IRegressionProblemData data) {
|
---|
184 | Name = data.Name;
|
---|
185 | Description = data.Description;
|
---|
186 | ProblemData = data;
|
---|
187 | }
|
---|
188 |
|
---|
189 | public IRegressionProblemData Export() {
|
---|
190 | return ProblemData;
|
---|
191 | }
|
---|
192 | #endregion
|
---|
193 | }
|
---|
194 | }
|
---|