Free cookie consent management tool by TermsFeed Policy Generator

source: branches/Scheduling/HeuristicLab.Problems.DataAnalysis.Regression/3.3/Symbolic/Analyzers/SymbolicRegressionModelQualityAnalyzer.cs @ 6955

Last change on this file since 6955 was 5863, checked in by mkommend, 14 years ago

#1418: Added NonDiscoverableType attribute to outdated analyzers.
File size: 15.9 KB
Line 
1#region License Information
2/* HeuristicLab
3 * Copyright (C) 2002-2011 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
22using System;
23using System.Collections.Generic;
24using System.Linq;
25using HeuristicLab.Analysis;
26using HeuristicLab.Common;
27using HeuristicLab.Core;
28using HeuristicLab.Data;
29using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
30using HeuristicLab.Operators;
31using HeuristicLab.Optimization;
32using HeuristicLab.Parameters;
33using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
34using HeuristicLab.PluginInfrastructure;
35using HeuristicLab.Problems.DataAnalysis.Evaluators;
36using HeuristicLab.Problems.DataAnalysis.Symbolic;
37
38namespace HeuristicLab.Problems.DataAnalysis.Regression.Symbolic.Analyzers {
39  /// <summary>
40  /// "An operator for analyzing the quality of symbolic regression solutions symbolic expression tree encoding."
41  /// </summary>
42  [Item("SymbolicRegressionModelQualityAnalyzer", "An operator for analyzing the quality of symbolic regression solutions symbolic expression tree encoding.")]
43  [StorableClass]
44  [NonDiscoverableType]
45  public sealed class SymbolicRegressionModelQualityAnalyzer : SingleSuccessorOperator, ISymbolicRegressionAnalyzer {
46    private const string SymbolicExpressionTreeInterpreterParameterName = "SymbolicExpressionTreeInterpreter";
47    private const string SymbolicExpressionTreeParameterName = "SymbolicExpressionTree";
48    private const string ProblemDataParameterName = "ProblemData";
49    private const string ResultsParameterName = "Results";
50
51    private const string TrainingMeanSquaredErrorQualityParameterName = "Mean squared error (training)";
52    private const string MinTrainingMeanSquaredErrorQualityParameterName = "Min mean squared error (training)";
53    private const string MaxTrainingMeanSquaredErrorQualityParameterName = "Max mean squared error (training)";
54    private const string AverageTrainingMeanSquaredErrorQualityParameterName = "Average mean squared error (training)";
55    private const string BestTrainingMeanSquaredErrorQualityParameterName = "Best mean squared error (training)";
56
57    private const string TrainingAverageRelativeErrorQualityParameterName = "Average relative error (training)";
58    private const string MinTrainingAverageRelativeErrorQualityParameterName = "Min average relative error (training)";
59    private const string MaxTrainingAverageRelativeErrorQualityParameterName = "Max average relative error (training)";
60    private const string AverageTrainingAverageRelativeErrorQualityParameterName = "Average average relative error (training)";
61    private const string BestTrainingAverageRelativeErrorQualityParameterName = "Best average relative error (training)";
62
63    private const string TrainingRSquaredQualityParameterName = "R² (training)";
64    private const string MinTrainingRSquaredQualityParameterName = "Min R² (training)";
65    private const string MaxTrainingRSquaredQualityParameterName = "Max R² (training)";
66    private const string AverageTrainingRSquaredQualityParameterName = "Average R² (training)";
67    private const string BestTrainingRSquaredQualityParameterName = "Best R² (training)";
68
69    private const string TestMeanSquaredErrorQualityParameterName = "Mean squared error (test)";
70    private const string MinTestMeanSquaredErrorQualityParameterName = "Min mean squared error (test)";
71    private const string MaxTestMeanSquaredErrorQualityParameterName = "Max mean squared error (test)";
72    private const string AverageTestMeanSquaredErrorQualityParameterName = "Average mean squared error (test)";
73    private const string BestTestMeanSquaredErrorQualityParameterName = "Best mean squared error (test)";
74
75    private const string TestAverageRelativeErrorQualityParameterName = "Average relative error (test)";
76    private const string MinTestAverageRelativeErrorQualityParameterName = "Min average relative error (test)";
77    private const string MaxTestAverageRelativeErrorQualityParameterName = "Max average relative error (test)";
78    private const string AverageTestAverageRelativeErrorQualityParameterName = "Average average relative error (test)";
79    private const string BestTestAverageRelativeErrorQualityParameterName = "Best average relative error (test)";
80
81    private const string TestRSquaredQualityParameterName = "R² (test)";
82    private const string MinTestRSquaredQualityParameterName = "Min R² (test)";
83    private const string MaxTestRSquaredQualityParameterName = "Max R² (test)";
84    private const string AverageTestRSquaredQualityParameterName = "Average R² (test)";
85    private const string BestTestRSquaredQualityParameterName = "Best R² (test)";
86
87    private const string RSquaredValuesParameterName = "R²";
88    private const string MeanSquaredErrorValuesParameterName = "Mean squared error";
89    private const string RelativeErrorValuesParameterName = "Average relative error";
90
91    private const string UpperEstimationLimitParameterName = "UpperEstimationLimit";
92    private const string LowerEstimationLimitParameterName = "LowerEstimationLimit";
93
94    #region parameter properties
95    public ScopeTreeLookupParameter<SymbolicExpressionTree> SymbolicExpressionTreeParameter {
96      get { return (ScopeTreeLookupParameter<SymbolicExpressionTree>)Parameters[SymbolicExpressionTreeParameterName]; }
97    }
98    public IValueLookupParameter<ISymbolicExpressionTreeInterpreter> SymbolicExpressionTreeInterpreterParameter {
99      get { return (IValueLookupParameter<ISymbolicExpressionTreeInterpreter>)Parameters[SymbolicExpressionTreeInterpreterParameterName]; }
100    }
101    public IValueLookupParameter<DataAnalysisProblemData> ProblemDataParameter {
102      get { return (IValueLookupParameter<DataAnalysisProblemData>)Parameters[ProblemDataParameterName]; }
103    }
104    public IValueLookupParameter<DoubleValue> UpperEstimationLimitParameter {
105      get { return (IValueLookupParameter<DoubleValue>)Parameters[UpperEstimationLimitParameterName]; }
106    }
107    public IValueLookupParameter<DoubleValue> LowerEstimationLimitParameter {
108      get { return (IValueLookupParameter<DoubleValue>)Parameters[LowerEstimationLimitParameterName]; }
109    }
110    public ILookupParameter<ResultCollection> ResultsParameter {
111      get { return (ILookupParameter<ResultCollection>)Parameters[ResultsParameterName]; }
112    }
113    #endregion
114    #region properties
115    public DoubleValue UpperEstimationLimit {
116      get { return UpperEstimationLimitParameter.ActualValue; }
117    }
118    public DoubleValue LowerEstimationLimit {
119      get { return LowerEstimationLimitParameter.ActualValue; }
120    }
121    #endregion
122
123    [StorableConstructor]
124    private SymbolicRegressionModelQualityAnalyzer(bool deserializing) : base(deserializing) { }
125    private SymbolicRegressionModelQualityAnalyzer(SymbolicRegressionModelQualityAnalyzer original, Cloner cloner) : base(original, cloner) { }
126    public SymbolicRegressionModelQualityAnalyzer()
127      : base() {
128      Parameters.Add(new ScopeTreeLookupParameter<SymbolicExpressionTree>(SymbolicExpressionTreeParameterName, "The symbolic expression trees to analyze."));
129      Parameters.Add(new ValueLookupParameter<ISymbolicExpressionTreeInterpreter>(SymbolicExpressionTreeInterpreterParameterName, "The interpreter that should be used to calculate the output values of the symbolic expression tree."));
130      Parameters.Add(new ValueLookupParameter<DataAnalysisProblemData>(ProblemDataParameterName, "The problem data containing the input varaibles for the symbolic regression problem."));
131      Parameters.Add(new ValueLookupParameter<DoubleValue>(UpperEstimationLimitParameterName, "The upper limit that should be used as cut off value for the output values of symbolic expression trees."));
132      Parameters.Add(new ValueLookupParameter<DoubleValue>(LowerEstimationLimitParameterName, "The lower limit that should be used as cut off value for the output values of symbolic expression trees."));
133      Parameters.Add(new ValueLookupParameter<DataTable>(MeanSquaredErrorValuesParameterName, "The data table to collect mean squared error values."));
134      Parameters.Add(new ValueLookupParameter<DataTable>(RSquaredValuesParameterName, "The data table to collect R² correlation coefficient values."));
135      Parameters.Add(new ValueLookupParameter<DataTable>(RelativeErrorValuesParameterName, "The data table to collect relative error values."));
136      Parameters.Add(new LookupParameter<ResultCollection>(ResultsParameterName, "The result collection where the best symbolic regression solution should be stored."));
137    }
138
139    public override IDeepCloneable Clone(Cloner cloner) {
140      return new SymbolicRegressionModelQualityAnalyzer(this, cloner);
141    }
142
143    public override IOperation Apply() {
144      double upperEstimationLimit = UpperEstimationLimit != null ? UpperEstimationLimit.Value : double.PositiveInfinity;
145      double lowerEstimationLimit = LowerEstimationLimit != null ? LowerEstimationLimit.Value : double.NegativeInfinity;
146      Analyze(SymbolicExpressionTreeParameter.ActualValue, SymbolicExpressionTreeInterpreterParameter.ActualValue,
147        upperEstimationLimit, lowerEstimationLimit, ProblemDataParameter.ActualValue,
148        ResultsParameter.ActualValue);
149      return base.Apply();
150    }
151
152    public static void Analyze(IEnumerable<SymbolicExpressionTree> trees, ISymbolicExpressionTreeInterpreter interpreter,
153      double upperEstimationLimit, double lowerEstimationLimit,
154      DataAnalysisProblemData problemData, ResultCollection results) {
155      int targetVariableIndex = problemData.Dataset.GetVariableIndex(problemData.TargetVariable.Value);
156      IEnumerable<double> originalTrainingValues = problemData.Dataset.GetEnumeratedVariableValues(targetVariableIndex, problemData.TrainingIndizes);
157      IEnumerable<double> originalTestValues = problemData.Dataset.GetEnumeratedVariableValues(targetVariableIndex, problemData.TestIndizes);
158      List<double> trainingMse = new List<double>();
159      List<double> trainingR2 = new List<double>();
160      List<double> trainingRelErr = new List<double>();
161      List<double> testMse = new List<double>();
162      List<double> testR2 = new List<double>();
163      List<double> testRelErr = new List<double>();
164
165      OnlineMeanSquaredErrorEvaluator mseEvaluator = new OnlineMeanSquaredErrorEvaluator();
166      OnlineMeanAbsolutePercentageErrorEvaluator relErrEvaluator = new OnlineMeanAbsolutePercentageErrorEvaluator();
167      OnlinePearsonsRSquaredEvaluator r2Evaluator = new OnlinePearsonsRSquaredEvaluator();
168
169      foreach (var tree in trees) {
170        #region training
171        var estimatedTrainingValues = interpreter.GetSymbolicExpressionTreeValues(tree, problemData.Dataset, problemData.TrainingIndizes);
172        mseEvaluator.Reset();
173        r2Evaluator.Reset();
174        relErrEvaluator.Reset();
175        var estimatedEnumerator = estimatedTrainingValues.GetEnumerator();
176        var originalEnumerator = originalTrainingValues.GetEnumerator();
177        while (estimatedEnumerator.MoveNext() & originalEnumerator.MoveNext()) {
178          double estimated = estimatedEnumerator.Current;
179          if (double.IsNaN(estimated)) estimated = upperEstimationLimit;
180          else estimated = Math.Min(upperEstimationLimit, Math.Max(lowerEstimationLimit, estimated));
181          mseEvaluator.Add(originalEnumerator.Current, estimated);
182          r2Evaluator.Add(originalEnumerator.Current, estimated);
183          relErrEvaluator.Add(originalEnumerator.Current, estimated);
184        }
185        if (estimatedEnumerator.MoveNext() || originalEnumerator.MoveNext()) {
186          throw new InvalidOperationException("Number of elements in estimated and original enumeration doesn't match.");
187        }
188        trainingMse.Add(mseEvaluator.MeanSquaredError);
189        trainingR2.Add(r2Evaluator.RSquared);
190        trainingRelErr.Add(relErrEvaluator.MeanAbsolutePercentageError);
191        #endregion
192        #region test
193        var estimatedTestValues = interpreter.GetSymbolicExpressionTreeValues(tree, problemData.Dataset, problemData.TestIndizes);
194
195        mseEvaluator.Reset();
196        r2Evaluator.Reset();
197        relErrEvaluator.Reset();
198        estimatedEnumerator = estimatedTestValues.GetEnumerator();
199        originalEnumerator = originalTestValues.GetEnumerator();
200        while (estimatedEnumerator.MoveNext() & originalEnumerator.MoveNext()) {
201          double estimated = estimatedEnumerator.Current;
202          if (double.IsNaN(estimated)) estimated = upperEstimationLimit;
203          else estimated = Math.Min(upperEstimationLimit, Math.Max(lowerEstimationLimit, estimated));
204          mseEvaluator.Add(originalEnumerator.Current, estimated);
205          r2Evaluator.Add(originalEnumerator.Current, estimated);
206          relErrEvaluator.Add(originalEnumerator.Current, estimated);
207        }
208        if (estimatedEnumerator.MoveNext() || originalEnumerator.MoveNext()) {
209          throw new InvalidOperationException("Number of elements in estimated and original enumeration doesn't match.");
210        }
211        testMse.Add(mseEvaluator.MeanSquaredError);
212        testR2.Add(r2Evaluator.RSquared);
213        testRelErr.Add(relErrEvaluator.MeanAbsolutePercentageError);
214        #endregion
215      }
216
217      AddResultTableValues(results, MeanSquaredErrorValuesParameterName, "mean squared error (training)", trainingMse.Min(), trainingMse.Average(), trainingMse.Max());
218      AddResultTableValues(results, MeanSquaredErrorValuesParameterName, "mean squared error (test)", testMse.Min(), testMse.Average(), testMse.Max());
219      AddResultTableValues(results, RelativeErrorValuesParameterName, "mean relative error (training)", trainingRelErr.Min(), trainingRelErr.Average(), trainingRelErr.Max());
220      AddResultTableValues(results, RelativeErrorValuesParameterName, "mean relative error (test)", testRelErr.Min(), testRelErr.Average(), testRelErr.Max());
221      AddResultTableValues(results, RSquaredValuesParameterName, "Pearson's R² (training)", trainingR2.Min(), trainingR2.Average(), trainingR2.Max());
222      AddResultTableValues(results, RSquaredValuesParameterName, "Pearson's R² (test)", testR2.Min(), testR2.Average(), testR2.Max());
223    }
224
225    private static void AddResultTableValues(ResultCollection results, string tableName, string valueName, double minValue, double avgValue, double maxValue) {
226      if (!results.ContainsKey(tableName)) {
227        results.Add(new Result(tableName, new DataTable(tableName)));
228      }
229      DataTable table = (DataTable)results[tableName].Value;
230      AddValue(table, minValue, "Min. " + valueName, string.Empty);
231      AddValue(table, avgValue, "Avg. " + valueName, string.Empty);
232      AddValue(table, maxValue, "Max. " + valueName, string.Empty);
233    }
234
235    private static void AddValue(DataTable table, double data, string name, string description) {
236      DataRow row;
237      table.Rows.TryGetValue(name, out row);
238      if (row == null) {
239        row = new DataRow(name, description);
240        row.Values.Add(data);
241        table.Rows.Add(row);
242      } else {
243        row.Values.Add(data);
244      }
245    }
246
247
248    private static void SetResultValue(ResultCollection results, string name, double value) {
249      if (results.ContainsKey(name))
250        results[name].Value = new DoubleValue(value);
251      else
252        results.Add(new Result(name, new DoubleValue(value)));
253    }
254  }
255}
Note: See TracBrowser for help on using the repository browser.