Free cookie consent management tool by TermsFeed Policy Generator

source: trunk/sources/HeuristicLab.Problems.DataAnalysis.Symbolic.Classification.Views/3.4/InteractiveSymbolicDiscriminantFunctionClassificationSolutionSimplifierView.cs @ 6803

Last change on this file since 6803 was 6803, checked in by mkommend, 11 years ago

#1479: Merged grammar editor branch into trunk.

File size: 7.4 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.Common;
26using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
27using HeuristicLab.Problems.DataAnalysis.Symbolic.Views;
28
29namespace HeuristicLab.Problems.DataAnalysis.Symbolic.Classification.Views {
30  public partial class InteractiveSymbolicDiscriminantFunctionClassificationSolutionSimplifierView : InteractiveSymbolicDataAnalysisSolutionSimplifierView {
31    private readonly ConstantTreeNode constantNode;
32    private readonly SymbolicExpressionTree tempTree;
33
34    public new SymbolicDiscriminantFunctionClassificationSolution Content {
35      get { return (SymbolicDiscriminantFunctionClassificationSolution)base.Content; }
36      set { base.Content = value; }
37    }
38
39    public InteractiveSymbolicDiscriminantFunctionClassificationSolutionSimplifierView()
40      : base() {
41      InitializeComponent();
42      this.Caption = "Interactive Classification Solution Simplifier";
43
44      constantNode = ((ConstantTreeNode)new Constant().CreateTreeNode());
45      ISymbolicExpressionTreeNode root = new ProgramRootSymbol().CreateTreeNode();
46      ISymbolicExpressionTreeNode start = new StartSymbol().CreateTreeNode();
47      root.AddSubtree(start);
48      tempTree = new SymbolicExpressionTree(root);
49    }
50
51    protected override void UpdateModel(ISymbolicExpressionTree tree) {
52      Content.Model = new SymbolicDiscriminantFunctionClassificationModel(tree, Content.Model.Interpreter);
53      // the default policy for setting thresholds in classification models is the accuarcy maximizing policy
54      // however for performance reasons we must use estimations of the normal distribution cut points as the thresholds
55      // here and in CalculateImpactValues as they are a lot faster to calculate
56      Content.SetClassDistibutionCutPointThresholds();
57    }
58
59    protected override Dictionary<ISymbolicExpressionTreeNode, double> CalculateReplacementValues(ISymbolicExpressionTree tree) {
60      Dictionary<ISymbolicExpressionTreeNode, double> replacementValues = new Dictionary<ISymbolicExpressionTreeNode, double>();
61      foreach (ISymbolicExpressionTreeNode node in tree.Root.GetSubtree(0).GetSubtree(0).IterateNodesPrefix()) {
62        replacementValues[node] = CalculateReplacementValue(node, tree);
63      }
64      return replacementValues;
65    }
66
67    protected override Dictionary<ISymbolicExpressionTreeNode, double> CalculateImpactValues(ISymbolicExpressionTree tree) {
68      var interpreter = Content.Model.Interpreter;
69      var dataset = Content.ProblemData.Dataset;
70      var rows = Content.ProblemData.TrainingIndizes;
71      string targetVariable = Content.ProblemData.TargetVariable;
72      Dictionary<ISymbolicExpressionTreeNode, double> impactValues = new Dictionary<ISymbolicExpressionTreeNode, double>();
73      List<ISymbolicExpressionTreeNode> nodes = tree.Root.GetSubtree(0).GetSubtree(0).IterateNodesPostfix().ToList();
74
75      var targetClassValues = dataset.GetDoubleValues(targetVariable, rows);
76      var originalOutput = interpreter.GetSymbolicExpressionTreeValues(tree, dataset, rows)
77        .LimitToRange(Content.Model.LowerEstimationLimit, Content.Model.UpperEstimationLimit)
78        .ToArray();
79      double[] classValues;
80      double[] thresholds;
81      // normal distribution cut points are used as thresholds here because they are a lot faster to calculate than the accuracy maximizing thresholds
82      NormalDistributionCutPointsThresholdCalculator.CalculateThresholds(Content.ProblemData, originalOutput, targetClassValues, out classValues, out thresholds);
83      var classifier = new SymbolicDiscriminantFunctionClassificationModel(tree, interpreter);
84      classifier.SetThresholdsAndClassValues(thresholds, classValues);
85      OnlineCalculatorError errorState;
86      double originalAccuracy = OnlineAccuracyCalculator.Calculate(targetClassValues, classifier.GetEstimatedClassValues(dataset, rows), out errorState);
87      if (errorState != OnlineCalculatorError.None) originalAccuracy = 0.0;
88
89      foreach (ISymbolicExpressionTreeNode node in nodes) {
90        var parent = node.Parent;
91        constantNode.Value = CalculateReplacementValue(node, tree);
92        ISymbolicExpressionTreeNode replacementNode = constantNode;
93        SwitchNode(parent, node, replacementNode);
94        var newOutput = interpreter.GetSymbolicExpressionTreeValues(tree, dataset, rows)
95          .LimitToRange(Content.Model.LowerEstimationLimit, Content.Model.UpperEstimationLimit)
96          .ToArray();
97        NormalDistributionCutPointsThresholdCalculator.CalculateThresholds(Content.ProblemData, newOutput, targetClassValues, out classValues, out thresholds);
98        classifier = new SymbolicDiscriminantFunctionClassificationModel(tree, interpreter);
99        classifier.SetThresholdsAndClassValues(thresholds, classValues);
100        double newAccuracy = OnlineAccuracyCalculator.Calculate(targetClassValues, classifier.GetEstimatedClassValues(dataset, rows), out errorState);
101        if (errorState != OnlineCalculatorError.None) newAccuracy = 0.0;
102
103        // impact = 0 if no change
104        // impact < 0 if new solution is better
105        // impact > 0 if new solution is worse
106        impactValues[node] = originalAccuracy - newAccuracy;
107        SwitchNode(parent, replacementNode, node);
108      }
109      return impactValues;
110    }
111
112    private double CalculateReplacementValue(ISymbolicExpressionTreeNode node, ISymbolicExpressionTree sourceTree) {
113      // remove old ADFs
114      while (tempTree.Root.SubtreeCount > 1) tempTree.Root.RemoveSubtree(1);
115      // clone ADFs of source tree
116      for (int i = 1; i < sourceTree.Root.SubtreeCount; i++) {
117        tempTree.Root.AddSubtree((ISymbolicExpressionTreeNode)sourceTree.Root.GetSubtree(i).Clone());
118      }
119      var start = tempTree.Root.GetSubtree(0);
120      while (start.SubtreeCount > 0) start.RemoveSubtree(0);
121      start.AddSubtree((ISymbolicExpressionTreeNode)node.Clone());
122      var interpreter = Content.Model.Interpreter;
123      var rows = Content.ProblemData.TrainingIndizes;
124      return interpreter.GetSymbolicExpressionTreeValues(tempTree, Content.ProblemData.Dataset, rows).Median();
125    }
126
127
128    private void SwitchNode(ISymbolicExpressionTreeNode root, ISymbolicExpressionTreeNode oldBranch, ISymbolicExpressionTreeNode newBranch) {
129      for (int i = 0; i < root.SubtreeCount; i++) {
130        if (root.GetSubtree(i) == oldBranch) {
131          root.RemoveSubtree(i);
132          root.InsertSubtree(i, newBranch);
133          return;
134        }
135      }
136    }
137
138    protected override void btnOptimizeConstants_Click(object sender, EventArgs e) {
139
140    }
141  }
142}
Note: See TracBrowser for help on using the repository browser.