Free cookie consent management tool by TermsFeed Policy Generator

source: trunk/sources/HeuristicLab.Problems.DataAnalysis.Regression/3.3/Symbolic/Analyzers/SymbolicRegressionModelQualityAnalyzer.cs @ 4054

Last change on this file since 4054 was 4054, checked in by mkommend, 14 years ago

improved Analyzers for SymoblicRegressionProblems (ticket #1074)
File size: 15.9 KB
Line 
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
22using System.Linq;
23using HeuristicLab.Common;
24using HeuristicLab.Core;
25using HeuristicLab.Data;
26using HeuristicLab.Operators;
27using HeuristicLab.Optimization;
28using HeuristicLab.Parameters;
29using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
30using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
31using HeuristicLab.Problems.DataAnalysis.Regression.Symbolic;
32using HeuristicLab.Problems.DataAnalysis.Symbolic;
33using System.Collections.Generic;
34using HeuristicLab.Problems.DataAnalysis.Symbolic.Symbols;
35using HeuristicLab.Problems.DataAnalysis;
36using HeuristicLab.Analysis;
37using HeuristicLab.Problems.DataAnalysis.Evaluators;
38using HeuristicLab.Optimization.Operators;
39using System;
40
41namespace HeuristicLab.Problems.DataAnalysis.Regression.Symbolic.Analyzers {
42  /// <summary>
43  /// "An operator for analyzing the quality of symbolic regression solutions symbolic expression tree encoding."
44  /// </summary>
45  [Item("SymbolicRegressionModelQualityAnalyzer", "An operator for analyzing the quality of symbolic regression solutions symbolic expression tree encoding.")]
46  [StorableClass]
47  public sealed class SymbolicRegressionModelQualityAnalyzer : SingleSuccessorOperator, ISymbolicRegressionAnalyzer {
48    private const string SymbolicExpressionTreeInterpreterParameterName = "SymbolicExpressionTreeInterpreter";
49    private const string SymbolicExpressionTreeParameterName = "SymbolicExpressionTree";
50    private const string ProblemDataParameterName = "ProblemData";
51    private const string ResultsParameterName = "Results";
52
53    private const string TrainingMeanSquaredErrorQualityParameterName = "Mean squared error (training)";
54    private const string MinTrainingMeanSquaredErrorQualityParameterName = "Min mean squared error (training)";
55    private const string MaxTrainingMeanSquaredErrorQualityParameterName = "Max mean squared error (training)";
56    private const string AverageTrainingMeanSquaredErrorQualityParameterName = "Average mean squared error (training)";
57    private const string BestTrainingMeanSquaredErrorQualityParameterName = "Best mean squared error (training)";
58
59    private const string TrainingAverageRelativeErrorQualityParameterName = "Average relative error (training)";
60    private const string MinTrainingAverageRelativeErrorQualityParameterName = "Min average relative error (training)";
61    private const string MaxTrainingAverageRelativeErrorQualityParameterName = "Max average relative error (training)";
62    private const string AverageTrainingAverageRelativeErrorQualityParameterName = "Average average relative error (training)";
63    private const string BestTrainingAverageRelativeErrorQualityParameterName = "Best average relative error (training)";
64
65    private const string TrainingRSquaredQualityParameterName = "R² (training)";
66    private const string MinTrainingRSquaredQualityParameterName = "Min R² (training)";
67    private const string MaxTrainingRSquaredQualityParameterName = "Max R² (training)";
68    private const string AverageTrainingRSquaredQualityParameterName = "Average R² (training)";
69    private const string BestTrainingRSquaredQualityParameterName = "Best R² (training)";
70
71    private const string TestMeanSquaredErrorQualityParameterName = "Mean squared error (test)";
72    private const string MinTestMeanSquaredErrorQualityParameterName = "Min mean squared error (test)";
73    private const string MaxTestMeanSquaredErrorQualityParameterName = "Max mean squared error (test)";
74    private const string AverageTestMeanSquaredErrorQualityParameterName = "Average mean squared error (test)";
75    private const string BestTestMeanSquaredErrorQualityParameterName = "Best mean squared error (test)";
76
77    private const string TestAverageRelativeErrorQualityParameterName = "Average relative error (test)";
78    private const string MinTestAverageRelativeErrorQualityParameterName = "Min average relative error (test)";
79    private const string MaxTestAverageRelativeErrorQualityParameterName = "Max average relative error (test)";
80    private const string AverageTestAverageRelativeErrorQualityParameterName = "Average average relative error (test)";
81    private const string BestTestAverageRelativeErrorQualityParameterName = "Best average relative error (test)";
82
83    private const string TestRSquaredQualityParameterName = "R² (test)";
84    private const string MinTestRSquaredQualityParameterName = "Min R² (test)";
85    private const string MaxTestRSquaredQualityParameterName = "Max R² (test)";
86    private const string AverageTestRSquaredQualityParameterName = "Average R² (test)";
87    private const string BestTestRSquaredQualityParameterName = "Best R² (test)";
88
89    private const string RSquaredValuesParameterName = "R²";
90    private const string MeanSquaredErrorValuesParameterName = "Mean squared error";
91    private const string RelativeErrorValuesParameterName = "Average relative error";
92
93    private const string UpperEstimationLimitParameterName = "UpperEstimationLimit";
94    private const string LowerEstimationLimitParameterName = "LowerEstimationLimit";
95
96    #region parameter properties
97    public ScopeTreeLookupParameter<SymbolicExpressionTree> SymbolicExpressionTreeParameter {
98      get { return (ScopeTreeLookupParameter<SymbolicExpressionTree>)Parameters[SymbolicExpressionTreeParameterName]; }
99    }
100    public IValueLookupParameter<ISymbolicExpressionTreeInterpreter> SymbolicExpressionTreeInterpreterParameter {
101      get { return (IValueLookupParameter<ISymbolicExpressionTreeInterpreter>)Parameters[SymbolicExpressionTreeInterpreterParameterName]; }
102    }
103    public IValueLookupParameter<DataAnalysisProblemData> ProblemDataParameter {
104      get { return (IValueLookupParameter<DataAnalysisProblemData>)Parameters[ProblemDataParameterName]; }
105    }
106    public IValueLookupParameter<DoubleValue> UpperEstimationLimitParameter {
107      get { return (IValueLookupParameter<DoubleValue>)Parameters[UpperEstimationLimitParameterName]; }
108    }
109    public IValueLookupParameter<DoubleValue> LowerEstimationLimitParameter {
110      get { return (IValueLookupParameter<DoubleValue>)Parameters[LowerEstimationLimitParameterName]; }
111    }
112    public ILookupParameter<ResultCollection> ResultsParameter {
113      get { return (ILookupParameter<ResultCollection>)Parameters[ResultsParameterName]; }
114    }
115    #endregion
116    #region properties
117    public DoubleValue UpperEstimationLimit {
118      get { return UpperEstimationLimitParameter.ActualValue; }
119    }
120    public DoubleValue LowerEstimationLimit {
121      get { return LowerEstimationLimitParameter.ActualValue; }
122    }
123    #endregion
124
125    public SymbolicRegressionModelQualityAnalyzer()
126      : base() {
127      Parameters.Add(new ScopeTreeLookupParameter<SymbolicExpressionTree>(SymbolicExpressionTreeParameterName, "The symbolic expression trees to analyze."));
128      Parameters.Add(new ValueLookupParameter<ISymbolicExpressionTreeInterpreter>(SymbolicExpressionTreeInterpreterParameterName, "The interpreter that should be used to calculate the output values of the symbolic expression tree."));
129      Parameters.Add(new ValueLookupParameter<DataAnalysisProblemData>(ProblemDataParameterName, "The problem data containing the input varaibles for the symbolic regression problem."));
130      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."));
131      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."));
132      Parameters.Add(new ValueLookupParameter<DataTable>(MeanSquaredErrorValuesParameterName, "The data table to collect mean squared error values."));
133      Parameters.Add(new ValueLookupParameter<DataTable>(RSquaredValuesParameterName, "The data table to collect R² correlation coefficient values."));
134      Parameters.Add(new ValueLookupParameter<DataTable>(RelativeErrorValuesParameterName, "The data table to collect relative error values."));
135      Parameters.Add(new LookupParameter<ResultCollection>(ResultsParameterName, "The result collection where the best symbolic regression solution should be stored."));
136    }
137
138    [StorableConstructor]
139    private SymbolicRegressionModelQualityAnalyzer(bool deserializing) : base() { }
140
141    public override IOperation Apply() {
142      Analyze(SymbolicExpressionTreeParameter.ActualValue, SymbolicExpressionTreeInterpreterParameter.ActualValue,
143        UpperEstimationLimit.Value, LowerEstimationLimit.Value, ProblemDataParameter.ActualValue,
144        ProblemDataParameter.ActualValue.TrainingSamplesStart.Value, ProblemDataParameter.ActualValue.TrainingSamplesEnd.Value,
145        ProblemDataParameter.ActualValue.TestSamplesStart.Value, ProblemDataParameter.ActualValue.TestSamplesEnd.Value,
146        ResultsParameter.ActualValue);
147      return base.Apply();
148    }
149
150    public static void Analyze(IEnumerable<SymbolicExpressionTree> trees, ISymbolicExpressionTreeInterpreter interpreter,
151      double upperEstimationLimit, double lowerEstimationLimit,
152      DataAnalysisProblemData problemData, int trainingStart, int trainingEnd, int testStart, int testEnd, ResultCollection results) {
153      int targetVariableIndex = problemData.Dataset.GetVariableIndex(problemData.TargetVariable.Value);
154      IEnumerable<double> originalTrainingValues = problemData.Dataset.GetEnumeratedVariableValues(targetVariableIndex, trainingStart, trainingEnd);
155      IEnumerable<double> originalTestValues = problemData.Dataset.GetEnumeratedVariableValues(targetVariableIndex, testStart, testEnd);
156      List<double> trainingMse = new List<double>();
157      List<double> trainingR2 = new List<double>();
158      List<double> trainingRelErr = new List<double>();
159      List<double> testMse = new List<double>();
160      List<double> testR2 = new List<double>();
161      List<double> testRelErr = new List<double>();
162
163      OnlineMeanSquaredErrorEvaluator mseEvaluator = new OnlineMeanSquaredErrorEvaluator();
164      OnlineMeanAbsolutePercentageErrorEvaluator relErrEvaluator = new OnlineMeanAbsolutePercentageErrorEvaluator();
165      OnlinePearsonsRSquaredEvaluator r2Evaluator = new OnlinePearsonsRSquaredEvaluator();
166
167      foreach (var tree in trees) {
168        #region training
169        var estimatedTrainingValues = interpreter.GetSymbolicExpressionTreeValues(tree, problemData.Dataset, Enumerable.Range(trainingStart, trainingEnd - trainingStart));
170        mseEvaluator.Reset();
171        r2Evaluator.Reset();
172        relErrEvaluator.Reset();
173        var estimatedEnumerator = estimatedTrainingValues.GetEnumerator();
174        var originalEnumerator = originalTrainingValues.GetEnumerator();
175        while (estimatedEnumerator.MoveNext() & originalEnumerator.MoveNext()) {
176          double estimated = estimatedEnumerator.Current;
177          if (double.IsNaN(estimated)) estimated = upperEstimationLimit;
178          else estimated = Math.Min(upperEstimationLimit, Math.Max(lowerEstimationLimit, estimated));
179          mseEvaluator.Add(originalEnumerator.Current, estimated);
180          r2Evaluator.Add(originalEnumerator.Current, estimated);
181          relErrEvaluator.Add(originalEnumerator.Current, estimated);
182        }
183        if (estimatedEnumerator.MoveNext() || originalEnumerator.MoveNext()) {
184          throw new InvalidOperationException("Number of elements in estimated and original enumeration doesn't match.");
185        }
186        trainingMse.Add(mseEvaluator.MeanSquaredError);
187        trainingR2.Add(r2Evaluator.RSquared);
188        trainingRelErr.Add(relErrEvaluator.MeanAbsolutePercentageError);
189        #endregion
190        #region test
191        var estimatedTestValues = interpreter.GetSymbolicExpressionTreeValues(tree, problemData.Dataset, Enumerable.Range(testStart, testEnd - testStart));
192
193        mseEvaluator.Reset();
194        r2Evaluator.Reset();
195        relErrEvaluator.Reset();
196        estimatedEnumerator = estimatedTestValues.GetEnumerator();
197        originalEnumerator = originalTestValues.GetEnumerator();
198        while (estimatedEnumerator.MoveNext() & originalEnumerator.MoveNext()) {
199          double estimated = estimatedEnumerator.Current;
200          if (double.IsNaN(estimated)) estimated = upperEstimationLimit;
201          else estimated = Math.Min(upperEstimationLimit, Math.Max(lowerEstimationLimit, estimated));
202          mseEvaluator.Add(originalEnumerator.Current, estimated);
203          r2Evaluator.Add(originalEnumerator.Current, estimated);
204          relErrEvaluator.Add(originalEnumerator.Current, estimated);
205        }
206        if (estimatedEnumerator.MoveNext() || originalEnumerator.MoveNext()) {
207          throw new InvalidOperationException("Number of elements in estimated and original enumeration doesn't match.");
208        }
209        testMse.Add(mseEvaluator.MeanSquaredError);
210        testR2.Add(r2Evaluator.RSquared);
211        testRelErr.Add(relErrEvaluator.MeanAbsolutePercentageError);
212        #endregion
213      }
214
215      AddResultTableValues(results, MeanSquaredErrorValuesParameterName, "mean squared error (training)", trainingMse.Min(), trainingMse.Average(), trainingMse.Max());
216      AddResultTableValues(results, MeanSquaredErrorValuesParameterName, "mean squared error (test)", testMse.Min(), testMse.Average(), testMse.Max());
217      AddResultTableValues(results, RelativeErrorValuesParameterName, "mean relative error (training)", trainingRelErr.Min(), trainingRelErr.Average(), trainingRelErr.Max());
218      AddResultTableValues(results, RelativeErrorValuesParameterName, "mean relative error (test)", testRelErr.Min(), testRelErr.Average(), testRelErr.Max());
219      AddResultTableValues(results, RSquaredValuesParameterName, "Pearson's R² (training)", trainingR2.Min(), trainingR2.Average(), trainingR2.Max());
220      AddResultTableValues(results, RSquaredValuesParameterName, "Pearson's R² (test)", testR2.Min(), testR2.Average(), testR2.Max());
221    }
222
223    private static void AddResultTableValues(ResultCollection results, string tableName, string valueName, double minValue, double avgValue, double maxValue) {
224      if (!results.ContainsKey(tableName)) {
225        results.Add(new Result(tableName, new DataTable(tableName)));
226      }
227      DataTable table = (DataTable)results[tableName].Value;
228      AddValue(table, minValue, "Min. " + valueName, string.Empty);
229      AddValue(table, avgValue, "Avg. " + valueName, string.Empty);
230      AddValue(table, maxValue, "Max. " + valueName, string.Empty);
231    }
232
233    private static void AddValue(DataTable table, double data, string name, string description) {
234      DataRow row;
235      table.Rows.TryGetValue(name, out row);
236      if (row == null) {
237        row = new DataRow(name, description);
238        row.Values.Add(data);
239        table.Rows.Add(row);
240      } else {
241        row.Values.Add(data);
242      }
243    }
244
245
246    private static void SetResultValue(ResultCollection results, string name, double value) {
247      if (results.ContainsKey(name))
248        results[name].Value = new DoubleValue(value);
249      else
250        results.Add(new Result(name, new DoubleValue(value)));
251    }
252  }
253}
Note: See TracBrowser for help on using the repository browser.