source: trunk/sources/HeuristicLab.Algorithms.DataAnalysis/3.4/GradientBoostedTrees/RegressionTreeModel.cs @ 13030

Last change on this file since 13030 was 13030, checked in by gkronber, 6 years ago

#2490: fixed problem with thread-safety in RegressionTreeModel using a different (local) caching tactic.

File size: 6.0 KB
Line 
1#region License Information
2/* HeuristicLab
3 * Copyright (C) 2002-2015 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
4 * and the BEACON Center for the Study of Evolution in Action.
5 *
6 * This file is part of HeuristicLab.
7 *
8 * HeuristicLab is free software: you can redistribute it and/or modify
9 * it under the terms of the GNU General Public License as published by
10 * the Free Software Foundation, either version 3 of the License, or
11 * (at your option) any later version.
12 *
13 * HeuristicLab is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
16 * GNU General Public License for more details.
17 *
18 * You should have received a copy of the GNU General Public License
19 * along with HeuristicLab. If not, see <http://www.gnu.org/licenses/>.
20 */
21#endregion
22
23using System;
24using System.Collections.Generic;
25using System.Collections.ObjectModel;
26using System.Globalization;
27using System.Linq;
28using HeuristicLab.Common;
29using HeuristicLab.Core;
30using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
31using HeuristicLab.Problems.DataAnalysis;
32
33namespace HeuristicLab.Algorithms.DataAnalysis {
34  [StorableClass]
35  [Item("RegressionTreeModel", "Represents a decision tree for regression.")]
36  public sealed class RegressionTreeModel : NamedItem, IRegressionModel {
37
38    // trees are represented as a flat array   
39    internal struct TreeNode {
40      public readonly static string NO_VARIABLE = null;
41
42      public TreeNode(string varName, double val, int leftIdx = -1, int rightIdx = -1)
43        : this() {
44        VarName = varName;
45        Val = val;
46        LeftIdx = leftIdx;
47        RightIdx = rightIdx;
48      }
49
50      public string VarName { get; private set; } // name of the variable for splitting or NO_VARIABLE if terminal node
51      public double Val { get; private set; } // threshold
52      public int LeftIdx { get; private set; }
53      public int RightIdx { get; private set; }
54
55      // necessary because the default implementation of GetHashCode for structs in .NET would only return the hashcode of val here
56      public override int GetHashCode() {
57        return LeftIdx ^ RightIdx ^ Val.GetHashCode();
58      }
59      // necessary because of GetHashCode override
60      public override bool Equals(object obj) {
61        if (obj is TreeNode) {
62          var other = (TreeNode)obj;
63          return Val.Equals(other.Val) &&
64            LeftIdx.Equals(other.LeftIdx) &&
65            RightIdx.Equals(other.RightIdx) &&
66            EqualStrings(VarName, other.VarName);
67        } else {
68          return false;
69        }
70      }
71
72      private bool EqualStrings(string a, string b) {
73        return (a == null && b == null) ||
74               (a != null && b != null && a.Equals(b));
75      }
76    }
77
78    // not storable!
79    private TreeNode[] tree;
80
81    [Storable]
82    // to prevent storing the references to data caches in nodes
83    // TODO seemingly it is bad (performance-wise) to persist tuples (tuples are used as keys in a dictionary)
84    private Tuple<string, double, int, int>[] SerializedTree {
85      get { return tree.Select(t => Tuple.Create(t.VarName, t.Val, t.LeftIdx, t.RightIdx)).ToArray(); }
86      set { this.tree = value.Select(t => new TreeNode(t.Item1, t.Item2, t.Item3, t.Item4)).ToArray(); }
87    }
88
89    [StorableConstructor]
90    private RegressionTreeModel(bool serializing) : base(serializing) { }
91    // cloning ctor
92    private RegressionTreeModel(RegressionTreeModel original, Cloner cloner)
93      : base(original, cloner) {
94      if (original.tree != null) {
95        this.tree = new TreeNode[original.tree.Length];
96        Array.Copy(original.tree, this.tree, this.tree.Length);
97      }
98    }
99
100    internal RegressionTreeModel(TreeNode[] tree)
101      : base("RegressionTreeModel", "Represents a decision tree for regression.") {
102      this.tree = tree;
103    }
104
105    private static double GetPredictionForRow(TreeNode[] t, ReadOnlyCollection<double>[] columnCache, int nodeIdx, int row) {
106      while (nodeIdx != -1) {
107        var node = t[nodeIdx];
108        if (node.VarName == TreeNode.NO_VARIABLE)
109          return node.Val;
110
111        if (columnCache[nodeIdx][row] <= node.Val)
112          nodeIdx = node.LeftIdx;
113        else
114          nodeIdx = node.RightIdx;
115      }
116      throw new InvalidOperationException("Invalid tree in RegressionTreeModel");
117    }
118
119    public override IDeepCloneable Clone(Cloner cloner) {
120      return new RegressionTreeModel(this, cloner);
121    }
122
123    public IEnumerable<double> GetEstimatedValues(IDataset ds, IEnumerable<int> rows) {
124      // lookup columns for variableNames in one pass over the tree to speed up evaluation later on
125      ReadOnlyCollection<double>[] columnCache = new ReadOnlyCollection<double>[tree.Length];
126
127      for (int i = 0; i < tree.Length; i++) {
128        if (tree[i].VarName != TreeNode.NO_VARIABLE) {
129          columnCache[i] = ds.GetReadOnlyDoubleValues(tree[i].VarName);
130        }
131      }
132      return rows.Select(r => GetPredictionForRow(tree, columnCache, 0, r));
133    }
134
135    public IRegressionSolution CreateRegressionSolution(IRegressionProblemData problemData) {
136      return new RegressionSolution(this, new RegressionProblemData(problemData));
137    }
138
139    // mainly for debugging
140    public override string ToString() {
141      return TreeToString(0, "");
142    }
143
144    private string TreeToString(int idx, string part) {
145      var n = tree[idx];
146      if (n.VarName == TreeNode.NO_VARIABLE) {
147        return string.Format(CultureInfo.InvariantCulture, "{0} -> {1:F}{2}", part, n.Val, Environment.NewLine);
148      } else {
149        return
150          TreeToString(n.LeftIdx, string.Format(CultureInfo.InvariantCulture, "{0}{1}{2} <= {3:F}", part, string.IsNullOrEmpty(part) ? "" : " and ", n.VarName, n.Val))
151        + TreeToString(n.RightIdx, string.Format(CultureInfo.InvariantCulture, "{0}{1}{2} >  {3:F}", part, string.IsNullOrEmpty(part) ? "" : " and ", n.VarName, n.Val));
152      }
153    }
154  }
155}
Note: See TracBrowser for help on using the repository browser.