[13865] | 1 | #region License Information
|
---|
| 2 | /* HeuristicLab
|
---|
| 3 | * Copyright (C) 2002-2015 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;
|
---|
[14310] | 23 | using System.Collections;
|
---|
[13865] | 24 | using System.Collections.Generic;
|
---|
| 25 | using System.Linq;
|
---|
[14310] | 26 | using HeuristicLab.Analysis;
|
---|
[13865] | 27 | using HeuristicLab.Common;
|
---|
| 28 | using HeuristicLab.Core;
|
---|
| 29 | using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
|
---|
| 30 | using HeuristicLab.Optimization;
|
---|
| 31 | using HeuristicLab.Parameters;
|
---|
| 32 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
---|
| 33 | using HeuristicLab.Problems.DataAnalysis;
|
---|
| 34 | using HeuristicLab.Problems.Instances;
|
---|
| 35 |
|
---|
| 36 |
|
---|
| 37 | namespace HeuristicLab.Problems.GeneticProgramming.GlucosePrediction {
|
---|
| 38 | [Item("Blood Glucose Forecast", "See MedGEC Workshop at GECCO 2016")]
|
---|
| 39 | [Creatable(CreatableAttribute.Categories.GeneticProgrammingProblems, Priority = 999)]
|
---|
| 40 | [StorableClass]
|
---|
| 41 | public sealed class Problem : SymbolicExpressionTreeProblem, IRegressionProblem, IProblemInstanceConsumer<IRegressionProblemData>, IProblemInstanceExporter<IRegressionProblemData> {
|
---|
| 42 |
|
---|
| 43 | #region parameter names
|
---|
| 44 | private const string ProblemDataParameterName = "ProblemData";
|
---|
| 45 | #endregion
|
---|
| 46 |
|
---|
| 47 | #region Parameter Properties
|
---|
| 48 | IParameter IDataAnalysisProblem.ProblemDataParameter { get { return ProblemDataParameter; } }
|
---|
| 49 |
|
---|
| 50 | public IValueParameter<IRegressionProblemData> ProblemDataParameter {
|
---|
| 51 | get { return (IValueParameter<IRegressionProblemData>)Parameters[ProblemDataParameterName]; }
|
---|
| 52 | }
|
---|
| 53 | #endregion
|
---|
| 54 |
|
---|
| 55 | #region Properties
|
---|
| 56 | public IRegressionProblemData ProblemData {
|
---|
| 57 | get { return ProblemDataParameter.Value; }
|
---|
| 58 | set { ProblemDataParameter.Value = value; }
|
---|
| 59 | }
|
---|
| 60 | IDataAnalysisProblemData IDataAnalysisProblem.ProblemData { get { return ProblemData; } }
|
---|
| 61 | #endregion
|
---|
| 62 |
|
---|
| 63 | public event EventHandler ProblemDataChanged;
|
---|
| 64 |
|
---|
| 65 | public override bool Maximization {
|
---|
[14310] | 66 | get { return true; }
|
---|
[13865] | 67 | }
|
---|
| 68 |
|
---|
| 69 | #region item cloning and persistence
|
---|
| 70 | // persistence
|
---|
| 71 | [StorableConstructor]
|
---|
| 72 | private Problem(bool deserializing) : base(deserializing) { }
|
---|
| 73 | [StorableHook(HookType.AfterDeserialization)]
|
---|
| 74 | private void AfterDeserialization() {
|
---|
| 75 | RegisterEventHandlers();
|
---|
| 76 | }
|
---|
| 77 |
|
---|
| 78 | // cloning
|
---|
| 79 | private Problem(Problem original, Cloner cloner)
|
---|
| 80 | : base(original, cloner) {
|
---|
| 81 | RegisterEventHandlers();
|
---|
| 82 | }
|
---|
| 83 | public override IDeepCloneable Clone(Cloner cloner) { return new Problem(this, cloner); }
|
---|
| 84 | #endregion
|
---|
| 85 |
|
---|
| 86 | public Problem()
|
---|
| 87 | : base() {
|
---|
| 88 | Parameters.Add(new ValueParameter<IRegressionProblemData>(ProblemDataParameterName, "The data for the glucose prediction problem", new RegressionProblemData()));
|
---|
| 89 |
|
---|
| 90 | var g = new SimpleSymbolicExpressionGrammar(); // empty grammar is replaced in UpdateGrammar()
|
---|
| 91 | base.Encoding = new SymbolicExpressionTreeEncoding(g, 100, 17);
|
---|
| 92 |
|
---|
| 93 | UpdateGrammar();
|
---|
| 94 | RegisterEventHandlers();
|
---|
| 95 | }
|
---|
| 96 |
|
---|
| 97 |
|
---|
| 98 | public override double Evaluate(ISymbolicExpressionTree tree, IRandom random) {
|
---|
| 99 | var problemData = ProblemData;
|
---|
| 100 | var rows = problemData.TrainingIndices.ToArray();
|
---|
| 101 | var target = problemData.Dataset.GetDoubleValues(problemData.TargetVariable, rows);
|
---|
[14310] | 102 | var predicted0 = Interpreter.Apply(tree.Root.GetSubtree(0).GetSubtree(0).GetSubtree(0), problemData.Dataset, rows);
|
---|
| 103 | var predicted1 = Interpreter.Apply(tree.Root.GetSubtree(0).GetSubtree(0).GetSubtree(1), problemData.Dataset, rows);
|
---|
| 104 | var predicted2 = Interpreter.Apply(tree.Root.GetSubtree(0).GetSubtree(0).GetSubtree(2), problemData.Dataset, rows);
|
---|
[13865] | 105 |
|
---|
[14310] | 106 | var pred0_rsq = Rsq(predicted0, target);
|
---|
| 107 | var pred1_rsq = Rsq(predicted1, target);
|
---|
| 108 | var pred2_rsq = Rsq(predicted2, target);
|
---|
| 109 | return pred0_rsq + pred1_rsq + pred2_rsq;
|
---|
| 110 | }
|
---|
| 111 |
|
---|
| 112 | private double Rsq(IEnumerable<double> predicted, IEnumerable<double> target) {
|
---|
[13865] | 113 | // only take predictions for which the target is not NaN
|
---|
| 114 | var selectedTuples = target.Zip(predicted, Tuple.Create).Where(t => !double.IsNaN(t.Item1)).ToArray();
|
---|
| 115 | target = selectedTuples.Select(t => t.Item1);
|
---|
| 116 | predicted = selectedTuples.Select(t => t.Item2);
|
---|
| 117 |
|
---|
| 118 | OnlineCalculatorError errorState;
|
---|
[14310] | 119 | var r = OnlinePearsonsRCalculator.Calculate(target, predicted, out errorState);
|
---|
| 120 | if (errorState != OnlineCalculatorError.None) r = 0;
|
---|
| 121 | return r * r;
|
---|
[13865] | 122 | }
|
---|
| 123 |
|
---|
[14310] | 124 | public override void Analyze(ISymbolicExpressionTree[] trees, double[] qualities, ResultCollection results,
|
---|
| 125 | IRandom random) {
|
---|
[13865] | 126 | base.Analyze(trees, qualities, results, random);
|
---|
| 127 |
|
---|
| 128 | if (!results.ContainsKey("Solution")) {
|
---|
| 129 | results.Add(new Result("Solution", typeof(IRegressionSolution)));
|
---|
| 130 | }
|
---|
[14310] | 131 | if (!results.ContainsKey("Terms")) {
|
---|
| 132 | results.Add(new Result("Terms", typeof(DataTable)));
|
---|
| 133 | }
|
---|
[13865] | 134 |
|
---|
| 135 | var bestTree = trees.First();
|
---|
| 136 | var bestQuality = qualities.First();
|
---|
| 137 | for (int i = 1; i < trees.Length; i++) {
|
---|
[14310] | 138 | if (qualities[i] > bestQuality) {
|
---|
[13865] | 139 | bestQuality = qualities[i];
|
---|
| 140 | bestTree = trees[i];
|
---|
| 141 | }
|
---|
| 142 | }
|
---|
| 143 |
|
---|
[14310] | 144 |
|
---|
[13865] | 145 | var clonedProblemData = (IRegressionProblemData)ProblemData.Clone();
|
---|
[14310] | 146 | var rows = clonedProblemData.TrainingIndices.ToArray();
|
---|
| 147 | var target = clonedProblemData.Dataset.GetDoubleValues(clonedProblemData.TargetVariable, rows).ToArray();
|
---|
| 148 | var predicted0 =
|
---|
| 149 | Interpreter.Apply(bestTree.Root.GetSubtree(0).GetSubtree(0).GetSubtree(0), clonedProblemData.Dataset, rows)
|
---|
| 150 | .ToArray();
|
---|
| 151 | var predicted1 =
|
---|
| 152 | Interpreter.Apply(bestTree.Root.GetSubtree(0).GetSubtree(0).GetSubtree(1), clonedProblemData.Dataset, rows)
|
---|
| 153 | .ToArray();
|
---|
| 154 | var predicted2 =
|
---|
| 155 | Interpreter.Apply(bestTree.Root.GetSubtree(0).GetSubtree(0).GetSubtree(2), clonedProblemData.Dataset, rows)
|
---|
| 156 | .ToArray();
|
---|
| 157 |
|
---|
| 158 | var termsTable = new HeuristicLab.Analysis.DataTable("Terms");
|
---|
| 159 | var r0 = new DataRow("GlucTerm", "GlucTerm", predicted0);
|
---|
| 160 | var r1 = new DataRow("InsTerm", "InsTerm", predicted1);
|
---|
| 161 | r1.VisualProperties.SecondYAxis = true;
|
---|
| 162 | var r2 = new DataRow("ChTerm", "ChTerm", predicted2);
|
---|
| 163 | r2.VisualProperties.SecondYAxis = true;
|
---|
| 164 | var r3 = new DataRow("Target", "Target", target);
|
---|
| 165 | termsTable.Rows.Add(r0);
|
---|
| 166 | termsTable.Rows.Add(r1);
|
---|
| 167 | termsTable.Rows.Add(r2);
|
---|
| 168 | termsTable.Rows.Add(r3);
|
---|
| 169 | results["Terms"].Value = termsTable;
|
---|
| 170 |
|
---|
| 171 |
|
---|
| 172 | var filteredPredicted0 = rows.Where(r => !double.IsNaN(target[r])).Select(r => predicted0[r]).ToArray();
|
---|
| 173 | var filteredPredicted1 = rows.Where(r => !double.IsNaN(target[r])).Select(r => predicted1[r]).ToArray();
|
---|
| 174 | var filteredPredicted2 = rows.Where(r => !double.IsNaN(target[r])).Select(r => predicted2[r]).ToArray();
|
---|
| 175 | var filteredTarget = target.Where(t => !double.IsNaN(t)).ToArray();
|
---|
| 176 |
|
---|
| 177 |
|
---|
| 178 | var ds = new ModifiableDataset(new string[] { "pred0", "pred1", "pred2", "target" },
|
---|
| 179 | new List<IList>
|
---|
| 180 | {
|
---|
| 181 | filteredPredicted0.ToList(),
|
---|
| 182 | filteredPredicted1.ToList(),
|
---|
| 183 | filteredPredicted2.ToList(),
|
---|
| 184 | filteredTarget.ToList()
|
---|
| 185 | });
|
---|
| 186 | var lrProbData = new RegressionProblemData(ds, new string[] { "pred0", "pred1", "pred2" }, "target");
|
---|
| 187 | lrProbData.TrainingPartition.Start = clonedProblemData.TrainingPartition.Start;
|
---|
| 188 | lrProbData.TrainingPartition.End = clonedProblemData.TrainingPartition.End;
|
---|
| 189 | lrProbData.TestPartition.Start = clonedProblemData.TestPartition.Start;
|
---|
| 190 | lrProbData.TestPartition.End = clonedProblemData.TestPartition.End;
|
---|
| 191 |
|
---|
| 192 | try {
|
---|
| 193 | double rmsError, cvRmsError;
|
---|
| 194 | var lrSolution = HeuristicLab.Algorithms.DataAnalysis.LinearRegression.CreateLinearRegressionSolution(
|
---|
| 195 | lrProbData, out rmsError, out cvRmsError);
|
---|
| 196 | results["Solution"].Value = lrSolution;
|
---|
| 197 | } catch (Exception) {
|
---|
| 198 | // ignore
|
---|
| 199 | }
|
---|
[13865] | 200 | }
|
---|
| 201 |
|
---|
| 202 | #region events
|
---|
| 203 | private void RegisterEventHandlers() {
|
---|
| 204 | ProblemDataParameter.ValueChanged += new EventHandler(ProblemDataParameter_ValueChanged);
|
---|
| 205 | if (ProblemDataParameter.Value != null) ProblemDataParameter.Value.Changed += new EventHandler(ProblemData_Changed);
|
---|
| 206 | }
|
---|
| 207 |
|
---|
| 208 | private void ProblemDataParameter_ValueChanged(object sender, EventArgs e) {
|
---|
| 209 | ProblemDataParameter.Value.Changed += new EventHandler(ProblemData_Changed);
|
---|
| 210 | OnProblemDataChanged();
|
---|
| 211 | OnReset();
|
---|
| 212 | }
|
---|
| 213 |
|
---|
| 214 | private void ProblemData_Changed(object sender, EventArgs e) {
|
---|
| 215 | OnReset();
|
---|
| 216 | }
|
---|
| 217 |
|
---|
| 218 | private void OnProblemDataChanged() {
|
---|
| 219 | UpdateGrammar();
|
---|
| 220 |
|
---|
| 221 | var handler = ProblemDataChanged;
|
---|
| 222 | if (handler != null) handler(this, EventArgs.Empty);
|
---|
| 223 | }
|
---|
| 224 |
|
---|
| 225 | private void UpdateGrammar() {
|
---|
| 226 | // whenever ProblemData is changed we create a new grammar with the necessary symbols
|
---|
| 227 | var g = new Grammar();
|
---|
| 228 | Encoding.Grammar = g;
|
---|
| 229 | }
|
---|
| 230 | #endregion
|
---|
| 231 |
|
---|
| 232 | #region Import & Export
|
---|
| 233 | public void Load(IRegressionProblemData data) {
|
---|
| 234 | Name = data.Name;
|
---|
| 235 | Description = data.Description;
|
---|
| 236 | ProblemData = data;
|
---|
| 237 | }
|
---|
| 238 |
|
---|
| 239 | public IRegressionProblemData Export() {
|
---|
| 240 | return ProblemData;
|
---|
| 241 | }
|
---|
| 242 | #endregion
|
---|
| 243 | }
|
---|
| 244 | }
|
---|