source: trunk/sources/HeuristicLab.Problems.DataAnalysis/3.4/Implementation/Classification/ClassificationSolutionBase.cs @ 13102

Last change on this file since 13102 was 13102, checked in by gkronber, 5 years ago

#1998:

  • changed namespace and name of view
  • calculate f1 score only for solutions for binary classification problems
File size: 9.9 KB
Line 
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
22using System.Collections.Generic;
23using System.Linq;
24using HeuristicLab.Common;
25using HeuristicLab.Data;
26using HeuristicLab.Optimization;
27using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
28using HeuristicLab.Problems.DataAnalysis.OnlineCalculators;
29
30namespace HeuristicLab.Problems.DataAnalysis {
31  [StorableClass]
32  public abstract class ClassificationSolutionBase : DataAnalysisSolution, IClassificationSolution {
33    private const string TrainingAccuracyResultName = "Accuracy (training)";
34    private const string TestAccuracyResultName = "Accuracy (test)";
35    private const string TrainingNormalizedGiniCoefficientResultName = "Normalized Gini Coefficient (training)";
36    private const string TestNormalizedGiniCoefficientResultName = "Normalized Gini Coefficient (test)";
37    private const string ClassificationPerformanceMeasuresResultName = "Classification Performance Measures";
38
39    public new IClassificationModel Model {
40      get { return (IClassificationModel)base.Model; }
41      protected set { base.Model = value; }
42    }
43
44    public new IClassificationProblemData ProblemData {
45      get { return (IClassificationProblemData)base.ProblemData; }
46      set { base.ProblemData = value; }
47    }
48
49    #region Results
50    public double TrainingAccuracy {
51      get { return ((DoubleValue)this[TrainingAccuracyResultName].Value).Value; }
52      private set { ((DoubleValue)this[TrainingAccuracyResultName].Value).Value = value; }
53    }
54    public double TestAccuracy {
55      get { return ((DoubleValue)this[TestAccuracyResultName].Value).Value; }
56      private set { ((DoubleValue)this[TestAccuracyResultName].Value).Value = value; }
57    }
58    public double TrainingNormalizedGiniCoefficient {
59      get { return ((DoubleValue)this[TrainingNormalizedGiniCoefficientResultName].Value).Value; }
60      protected set { ((DoubleValue)this[TrainingNormalizedGiniCoefficientResultName].Value).Value = value; }
61    }
62    public double TestNormalizedGiniCoefficient {
63      get { return ((DoubleValue)this[TestNormalizedGiniCoefficientResultName].Value).Value; }
64      protected set { ((DoubleValue)this[TestNormalizedGiniCoefficientResultName].Value).Value = value; }
65    }
66    public ClassificationPerformanceMeasuresResultCollection ClassificationPerformanceMeasures {
67      get { return ((ClassificationPerformanceMeasuresResultCollection)this[ClassificationPerformanceMeasuresResultName].Value); }
68      protected set { (this[ClassificationPerformanceMeasuresResultName].Value) = value; }
69    }
70    #endregion
71
72    [StorableConstructor]
73    protected ClassificationSolutionBase(bool deserializing) : base(deserializing) { }
74    protected ClassificationSolutionBase(ClassificationSolutionBase original, Cloner cloner)
75      : base(original, cloner) {
76    }
77    protected ClassificationSolutionBase(IClassificationModel model, IClassificationProblemData problemData)
78      : base(model, problemData) {
79      Add(new Result(TrainingAccuracyResultName, "Accuracy of the model on the training partition (percentage of correctly classified instances).", new PercentValue()));
80      Add(new Result(TestAccuracyResultName, "Accuracy of the model on the test partition (percentage of correctly classified instances).", new PercentValue()));
81      Add(new Result(TrainingNormalizedGiniCoefficientResultName, "Normalized Gini coefficient of the model on the training partition.", new DoubleValue()));
82      Add(new Result(TestNormalizedGiniCoefficientResultName, "Normalized Gini coefficient of the model on the test partition.", new DoubleValue()));
83      Add(new Result(ClassificationPerformanceMeasuresResultName, @"Classification performance measures.\n
84                              In a multiclass classification all misclassifications of the negative class will be treated as true negatives except on positive class estimations.",
85                            new ClassificationPerformanceMeasuresResultCollection()));
86    }
87
88    [StorableHook(HookType.AfterDeserialization)]
89    private void AfterDeserialization() {
90      if (!this.ContainsKey(TrainingNormalizedGiniCoefficientResultName))
91        Add(new Result(TrainingNormalizedGiniCoefficientResultName, "Normalized Gini coefficient of the model on the training partition.", new DoubleValue()));
92      if (!this.ContainsKey(TestNormalizedGiniCoefficientResultName))
93        Add(new Result(TestNormalizedGiniCoefficientResultName, "Normalized Gini coefficient of the model on the test partition.", new DoubleValue()));
94      if (!this.ContainsKey(ClassificationPerformanceMeasuresResultName)) {
95        Add(new Result(ClassificationPerformanceMeasuresResultName, @"Classification performance measures.\n
96                              In a multiclass classification all misclassifications of the negative class will be treated as true negatives except on positive class estimations.",
97                              new ClassificationPerformanceMeasuresResultCollection()));
98        CalculateClassificationResults();
99      }
100    }
101
102    protected void CalculateClassificationResults() {
103      double[] estimatedTrainingClassValues = EstimatedTrainingClassValues.ToArray(); // cache values
104      double[] originalTrainingClassValues = ProblemData.Dataset.GetDoubleValues(ProblemData.TargetVariable, ProblemData.TrainingIndices).ToArray();
105
106      double[] estimatedTestClassValues = EstimatedTestClassValues.ToArray(); // cache values
107      double[] originalTestClassValues = ProblemData.Dataset.GetDoubleValues(ProblemData.TargetVariable, ProblemData.TestIndices).ToArray();
108
109      var positiveClassName = ProblemData.PositiveClass;
110      double positiveClassValue = ProblemData.GetClassValue(positiveClassName);
111      ClassificationPerformanceMeasuresCalculator trainingPerformanceCalculator = new ClassificationPerformanceMeasuresCalculator(positiveClassName, positiveClassValue);
112      ClassificationPerformanceMeasuresCalculator testPerformanceCalculator = new ClassificationPerformanceMeasuresCalculator(positiveClassName, positiveClassValue);
113
114      OnlineCalculatorError errorState;
115      double trainingAccuracy = OnlineAccuracyCalculator.Calculate(originalTrainingClassValues, estimatedTrainingClassValues, out errorState);
116      if (errorState != OnlineCalculatorError.None) trainingAccuracy = double.NaN;
117      double testAccuracy = OnlineAccuracyCalculator.Calculate(originalTestClassValues, estimatedTestClassValues, out errorState);
118      if (errorState != OnlineCalculatorError.None) testAccuracy = double.NaN;
119
120      TrainingAccuracy = trainingAccuracy;
121      TestAccuracy = testAccuracy;
122
123      double trainingNormalizedGini = NormalizedGiniCalculator.Calculate(originalTrainingClassValues, estimatedTrainingClassValues, out errorState);
124      if (errorState != OnlineCalculatorError.None) trainingNormalizedGini = double.NaN;
125      double testNormalizedGini = NormalizedGiniCalculator.Calculate(originalTestClassValues, estimatedTestClassValues, out errorState);
126      if (errorState != OnlineCalculatorError.None) testNormalizedGini = double.NaN;
127
128      TrainingNormalizedGiniCoefficient = trainingNormalizedGini;
129      TestNormalizedGiniCoefficient = testNormalizedGini;
130
131      trainingPerformanceCalculator.Calculate(originalTrainingClassValues, estimatedTrainingClassValues);
132      if (trainingPerformanceCalculator.ErrorState == OnlineCalculatorError.None)
133        ClassificationPerformanceMeasures.SetTrainingResults(trainingPerformanceCalculator);
134
135      testPerformanceCalculator.Calculate(originalTestClassValues, estimatedTestClassValues);
136      if (testPerformanceCalculator.ErrorState == OnlineCalculatorError.None)
137        ClassificationPerformanceMeasures.SetTestResults(testPerformanceCalculator);
138
139      if (ProblemData.Classes == 2) {
140        var f1Training = FOneScoreCalculator.Calculate(originalTrainingClassValues, estimatedTrainingClassValues, out errorState);
141        if (errorState == OnlineCalculatorError.None) ClassificationPerformanceMeasures.TrainingF1Score = f1Training;
142        var f1Test = FOneScoreCalculator.Calculate(originalTestClassValues, estimatedTestClassValues, out errorState);
143        if (errorState == OnlineCalculatorError.None) ClassificationPerformanceMeasures.TestF1Score = f1Test;
144      }
145
146      var mccTraining = MatthewsCorrelationCoefficientCalculator.Calculate(originalTrainingClassValues, estimatedTrainingClassValues, out errorState);
147      if (errorState == OnlineCalculatorError.None) ClassificationPerformanceMeasures.TrainingMatthewsCorrelation = mccTraining;
148      var mccTest = MatthewsCorrelationCoefficientCalculator.Calculate(originalTestClassValues, estimatedTestClassValues, out errorState);
149      if (errorState == OnlineCalculatorError.None) ClassificationPerformanceMeasures.TestMatthewsCorrelation = mccTest;
150    }
151
152    public abstract IEnumerable<double> EstimatedClassValues { get; }
153    public abstract IEnumerable<double> EstimatedTrainingClassValues { get; }
154    public abstract IEnumerable<double> EstimatedTestClassValues { get; }
155
156    public abstract IEnumerable<double> GetEstimatedClassValues(IEnumerable<int> rows);
157
158    protected override void RecalculateResults() {
159      CalculateClassificationResults();
160    }
161  }
162}
Note: See TracBrowser for help on using the repository browser.