Free cookie consent management tool by TermsFeed Policy Generator

source: trunk/sources/HeuristicLab.Problems.DataAnalysis.Symbolic.Classification/3.4/SymbolicDiscriminantFunctionClassificationSolution.cs @ 5809

Last change on this file since 5809 was 5809, checked in by mkommend, 13 years ago

#1418: Reintegrated branch into trunk.

File size: 6.7 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.Collections.Generic;
23using System.Linq;
24using HeuristicLab.Common;
25using HeuristicLab.Core;
26using HeuristicLab.Data;
27using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
28using HeuristicLab.Operators;
29using HeuristicLab.Parameters;
30using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
31using HeuristicLab.Optimization;
32using System;
33
34namespace HeuristicLab.Problems.DataAnalysis.Symbolic.Classification {
35  /// <summary>
36  /// Represents a symbolic classification solution (model + data) and attributes of the solution like accuracy and complexity
37  /// </summary>
38  [StorableClass]
39  [Item(Name = "SymbolicDiscriminantFunctionClassificationSolution", Description = "Represents a symbolic classification solution (model + data) and attributes of the solution like accuracy and complexity.")]
40  public sealed class SymbolicDiscriminantFunctionClassificationSolution : DiscriminantFunctionClassificationSolution, ISymbolicClassificationSolution {
41    private const string ModelLengthResultName = "ModelLength";
42    private const string ModelDepthResultName = "ModelDepth";
43
44    public new ISymbolicDiscriminantFunctionClassificationModel Model {
45      get { return (ISymbolicDiscriminantFunctionClassificationModel)base.Model; }
46      set { base.Model = value; }
47    }
48
49    ISymbolicClassificationModel ISymbolicClassificationSolution.Model {
50      get { return Model; }
51    }
52
53    ISymbolicDataAnalysisModel ISymbolicDataAnalysisSolution.Model {
54      get { return Model; }
55    }
56    public int ModelLength {
57      get { return ((IntValue)this[ModelLengthResultName].Value).Value; }
58      private set { ((IntValue)this[ModelLengthResultName].Value).Value = value; }
59    }
60
61    public int ModelDepth {
62      get { return ((IntValue)this[ModelDepthResultName].Value).Value; }
63      private set { ((IntValue)this[ModelDepthResultName].Value).Value = value; }
64    }
65    [StorableConstructor]
66    private SymbolicDiscriminantFunctionClassificationSolution(bool deserializing) : base(deserializing) { }
67    private SymbolicDiscriminantFunctionClassificationSolution(SymbolicDiscriminantFunctionClassificationSolution original, Cloner cloner)
68      : base(original, cloner) {
69    }
70    public SymbolicDiscriminantFunctionClassificationSolution(ISymbolicDiscriminantFunctionClassificationModel model, IClassificationProblemData problemData)
71      : base(model, problemData) {
72      Add(new Result(ModelLengthResultName, "Length of the symbolic classification model.", new IntValue()));
73      Add(new Result(ModelDepthResultName, "Depth of the symbolic classification model.", new IntValue()));
74      RecalculateResults();
75    }
76
77    public override IDeepCloneable Clone(Cloner cloner) {
78      return new SymbolicDiscriminantFunctionClassificationSolution(this, cloner);
79    }
80
81    protected override void OnModelChanged(EventArgs e) {
82      base.OnModelChanged(e);
83      RecalculateResults();
84    }
85
86    private new void RecalculateResults() {
87      ModelLength = Model.SymbolicExpressionTree.Length;
88      ModelDepth = Model.SymbolicExpressionTree.Depth;
89    }
90
91    public void ScaleModel() {
92      var dataset = ProblemData.Dataset;
93      var targetVariable = ProblemData.TargetVariable;
94      var rows = ProblemData.TrainingIndizes;
95      var estimatedValues = GetEstimatedValues(rows);
96      var targetValues = dataset.GetEnumeratedVariableValues(targetVariable, rows);
97      double alpha;
98      double beta;
99      OnlineLinearScalingParameterCalculator.Calculate(estimatedValues, targetValues, out alpha, out beta);
100
101      ConstantTreeNode alphaTreeNode = null;
102      ConstantTreeNode betaTreeNode = null;
103      // check if model has been scaled previously by analyzing the structure of the tree
104      var startNode = Model.SymbolicExpressionTree.Root.GetSubtree(0);
105      if (startNode.GetSubtree(0).Symbol is Addition) {
106        var addNode = startNode.GetSubtree(0);
107        if (addNode.SubtreesCount == 2 && addNode.GetSubtree(0).Symbol is Multiplication && addNode.GetSubtree(1).Symbol is Constant) {
108          alphaTreeNode = addNode.GetSubtree(1) as ConstantTreeNode;
109          var mulNode = addNode.GetSubtree(0);
110          if (mulNode.SubtreesCount == 2 && mulNode.GetSubtree(1).Symbol is Constant) {
111            betaTreeNode = mulNode.GetSubtree(1) as ConstantTreeNode;
112          }
113        }
114      }
115      // if tree structure matches the structure necessary for linear scaling then reuse the existing tree nodes
116      if (alphaTreeNode != null && betaTreeNode != null) {
117        betaTreeNode.Value *= beta;
118        alphaTreeNode.Value *= beta;
119        alphaTreeNode.Value += alpha;
120      } else {
121        var mainBranch = startNode.GetSubtree(0);
122        startNode.RemoveSubtree(0);
123        var scaledMainBranch = MakeSum(MakeProduct(beta, mainBranch), alpha);
124        startNode.AddSubtree(scaledMainBranch);
125      }
126
127      OnModelChanged(EventArgs.Empty);
128    }
129
130    private static ISymbolicExpressionTreeNode MakeSum(ISymbolicExpressionTreeNode treeNode, double alpha) {
131      if (alpha.IsAlmost(0.0)) {
132        return treeNode;
133      } else {
134        var node = (new Addition()).CreateTreeNode();
135        var alphaConst = MakeConstant(alpha);
136        node.AddSubtree(treeNode);
137        node.AddSubtree(alphaConst);
138        return node;
139      }
140    }
141
142    private static ISymbolicExpressionTreeNode MakeProduct(double beta, ISymbolicExpressionTreeNode treeNode) {
143      if (beta.IsAlmost(1.0)) {
144        return treeNode;
145      } else {
146        var node = (new Multiplication()).CreateTreeNode();
147        var betaConst = MakeConstant(beta);
148        node.AddSubtree(treeNode);
149        node.AddSubtree(betaConst);
150        return node;
151      }
152    }
153
154    private static ISymbolicExpressionTreeNode MakeConstant(double c) {
155      var node = (ConstantTreeNode)(new Constant()).CreateTreeNode();
156      node.Value = c;
157      return node;
158    }
159  }
160}
Note: See TracBrowser for help on using the repository browser.