source: branches/2925_AutoDiffForDynamicalModels/HeuristicLab.Problems.DataAnalysis/3.4/Implementation/Classification/DiscriminantFunctionClassificationModel.cs @ 16892

Last change on this file since 16892 was 16892, checked in by gkronber, 5 months ago

#2925 merged r16661:16890 from trunk to branch

File size: 6.4 KB
RevLine 
[5649]1#region License Information
2/* HeuristicLab
[16662]3 * Copyright (C) 2002-2019 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
[5649]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
[5777]22using System;
[5649]23using System.Collections.Generic;
24using System.Linq;
[16892]25using HEAL.Attic;
[5649]26using HeuristicLab.Common;
27using HeuristicLab.Core;
28
29namespace HeuristicLab.Problems.DataAnalysis {
30  /// <summary>
31  /// Represents discriminant function classification data analysis models.
32  /// </summary>
[16662]33  [StorableType("E7A8648D-C938-499F-A712-185542095708")]
[5649]34  [Item("DiscriminantFunctionClassificationModel", "Represents a classification model that uses a discriminant function and classification thresholds.")]
[13941]35  public class DiscriminantFunctionClassificationModel : ClassificationModel, IDiscriminantFunctionClassificationModel {
36    public override IEnumerable<string> VariablesUsedForPrediction {
[13921]37      get { return model.VariablesUsedForPrediction; }
38    }
39
[5649]40    [Storable]
41    private IRegressionModel model;
[8679]42    public IRegressionModel Model {
43      get { return model; }
44      private set { model = value; }
45    }
[5736]46
[5649]47    [Storable]
48    private double[] classValues;
[5678]49    public IEnumerable<double> ClassValues {
50      get { return (double[])classValues.Clone(); }
[5736]51      private set { classValues = value.ToArray(); }
[5678]52    }
[5736]53
[5678]54    [Storable]
55    private double[] thresholds;
56    public IEnumerable<double> Thresholds {
57      get { return (IEnumerable<double>)thresholds.Clone(); }
[5736]58      private set { thresholds = value.ToArray(); }
[5678]59    }
[5649]60
[8623]61    private IDiscriminantFunctionThresholdCalculator thresholdCalculator;
62    [Storable]
63    public IDiscriminantFunctionThresholdCalculator ThresholdCalculator {
64      get { return thresholdCalculator; }
65      private set { thresholdCalculator = value; }
66    }
[5678]67
[8623]68
[5649]69    [StorableConstructor]
[16662]70    protected DiscriminantFunctionClassificationModel(StorableConstructorFlag _) : base(_) { }
[5649]71    protected DiscriminantFunctionClassificationModel(DiscriminantFunctionClassificationModel original, Cloner cloner)
72      : base(original, cloner) {
73      model = cloner.Clone(original.model);
74      classValues = (double[])original.classValues.Clone();
[5678]75      thresholds = (double[])original.thresholds.Clone();
[5649]76    }
[5736]77
[8623]78    public DiscriminantFunctionClassificationModel(IRegressionModel model, IDiscriminantFunctionThresholdCalculator thresholdCalculator)
[13941]79      : base(model.TargetVariable) {
[5649]80      this.name = ItemName;
81      this.description = ItemDescription;
[13941]82
[5649]83      this.model = model;
[8623]84      this.classValues = new double[0];
85      this.thresholds = new double[0];
86      this.thresholdCalculator = thresholdCalculator;
[5649]87    }
88
[8623]89    [StorableHook(HookType.AfterDeserialization)]
90    private void AfterDeserialization() {
91      if (ThresholdCalculator == null) ThresholdCalculator = new AccuracyMaximizationThresholdCalculator();
92    }
93
[8679]94    public override IDeepCloneable Clone(Cloner cloner) {
95      return new DiscriminantFunctionClassificationModel(this, cloner);
96    }
97
[5736]98    public void SetThresholdsAndClassValues(IEnumerable<double> thresholds, IEnumerable<double> classValues) {
99      var classValuesArr = classValues.ToArray();
100      var thresholdsArr = thresholds.ToArray();
101      if (thresholdsArr.Length != classValuesArr.Length) throw new ArgumentException();
102
103      this.classValues = classValuesArr;
104      this.thresholds = thresholdsArr;
105      OnThresholdsChanged(EventArgs.Empty);
106    }
107
[8623]108    public virtual void RecalculateModelParameters(IClassificationProblemData problemData, IEnumerable<int> rows) {
109      double[] classValues;
110      double[] thresholds;
111      var targetClassValues = problemData.Dataset.GetDoubleValues(problemData.TargetVariable, rows);
112      var estimatedTrainingValues = GetEstimatedValues(problemData.Dataset, rows);
113      thresholdCalculator.Calculate(problemData, estimatedTrainingValues, targetClassValues, out classValues, out thresholds);
114      SetThresholdsAndClassValues(thresholds, classValues);
115    }
116
117
[12509]118    public IEnumerable<double> GetEstimatedValues(IDataset dataset, IEnumerable<int> rows) {
[5649]119      return model.GetEstimatedValues(dataset, rows);
120    }
121
[13941]122    public override IEnumerable<double> GetEstimatedClassValues(IDataset dataset, IEnumerable<int> rows) {
[16892]123      var estimatedValues = GetEstimatedValues(dataset, rows);
124      return GetEstimatedClassValues(estimatedValues);
125    }
126
127    public virtual IEnumerable<double> GetEstimatedClassValues(IEnumerable<double> estimatedValues) {
[8623]128      if (!Thresholds.Any() && !ClassValues.Any()) throw new ArgumentException("No thresholds and class values were set for the current classification model.");
[16892]129      foreach (var x in estimatedValues) {
[5649]130        int classIndex = 0;
[5678]131        // find first threshold value which is larger than x => class index = threshold index + 1
[5649]132        for (int i = 0; i < thresholds.Length; i++) {
133          if (x > thresholds[i]) classIndex++;
134          else break;
135        }
[5736]136        yield return classValues.ElementAt(classIndex - 1);
[5649]137      }
138    }
[16892]139
[5678]140    #region events
141    public event EventHandler ThresholdsChanged;
142    protected virtual void OnThresholdsChanged(EventArgs e) {
143      var listener = ThresholdsChanged;
144      if (listener != null) listener(this, e);
145    }
[5649]146    #endregion
[6604]147
[13941]148    public override IClassificationSolution CreateClassificationSolution(IClassificationProblemData problemData) {
149      return CreateDiscriminantFunctionClassificationSolution(problemData);
150    }
151    public virtual IDiscriminantFunctionClassificationSolution CreateDiscriminantFunctionClassificationSolution(
152      IClassificationProblemData problemData) {
[8857]153      return new DiscriminantFunctionClassificationSolution(this, new ClassificationProblemData(problemData));
[8679]154    }
[5649]155  }
156}
Note: See TracBrowser for help on using the repository browser.