[12590] | 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 |
|
---|
| 23 | using System;
|
---|
[12332] | 24 | using System.Collections.Generic;
|
---|
| 25 | using System.Linq;
|
---|
| 26 | using HeuristicLab.Common;
|
---|
| 27 | using HeuristicLab.Core;
|
---|
| 28 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
---|
| 29 | using HeuristicLab.Problems.DataAnalysis;
|
---|
| 30 |
|
---|
[12590] | 31 | namespace HeuristicLab.Algorithms.DataAnalysis {
|
---|
[12332] | 32 | [StorableClass]
|
---|
[12372] | 33 | [Item("Gradient boosted tree model", "")]
|
---|
[12590] | 34 | // this is essentially a collection of weighted regression models
|
---|
[13184] | 35 | public sealed class GradientBoostedTreesModel : NamedItem, IGradientBoostedTreesModel {
|
---|
| 36 | // BackwardsCompatibility3.4 for allowing deserialization & serialization of old models
|
---|
| 37 | #region Backwards compatible code, remove with 3.5
|
---|
| 38 | private bool isCompatibilityLoaded = false; // only set to true if the model is deserialized from the old format, needed to make sure that information is serialized again if it was loaded from the old format
|
---|
| 39 |
|
---|
| 40 | [Storable(Name = "models")]
|
---|
| 41 | private IList<IRegressionModel> __persistedModels {
|
---|
| 42 | set {
|
---|
| 43 | this.isCompatibilityLoaded = true;
|
---|
| 44 | this.models.Clear();
|
---|
| 45 | foreach (var m in value) this.models.Add(m);
|
---|
| 46 | }
|
---|
| 47 | get { if (this.isCompatibilityLoaded) return models; else return null; }
|
---|
| 48 | }
|
---|
| 49 | [Storable(Name = "weights")]
|
---|
| 50 | private IList<double> __persistedWeights {
|
---|
| 51 | set {
|
---|
| 52 | this.isCompatibilityLoaded = true;
|
---|
| 53 | this.weights.Clear();
|
---|
| 54 | foreach (var w in value) this.weights.Add(w);
|
---|
| 55 | }
|
---|
| 56 | get { if (this.isCompatibilityLoaded) return weights; else return null; }
|
---|
| 57 | }
|
---|
| 58 | #endregion
|
---|
| 59 |
|
---|
[12332] | 60 | private readonly IList<IRegressionModel> models;
|
---|
[12372] | 61 | public IEnumerable<IRegressionModel> Models { get { return models; } }
|
---|
| 62 |
|
---|
[12332] | 63 | private readonly IList<double> weights;
|
---|
[12372] | 64 | public IEnumerable<double> Weights { get { return weights; } }
|
---|
[12332] | 65 |
|
---|
| 66 | [StorableConstructor]
|
---|
[13184] | 67 | private GradientBoostedTreesModel(bool deserializing)
|
---|
| 68 | : base(deserializing) {
|
---|
| 69 | models = new List<IRegressionModel>();
|
---|
| 70 | weights = new List<double>();
|
---|
| 71 | }
|
---|
[12332] | 72 | private GradientBoostedTreesModel(GradientBoostedTreesModel original, Cloner cloner)
|
---|
| 73 | : base(original, cloner) {
|
---|
| 74 | this.weights = new List<double>(original.weights);
|
---|
| 75 | this.models = new List<IRegressionModel>(original.models.Select(m => cloner.Clone(m)));
|
---|
[13184] | 76 | this.isCompatibilityLoaded = original.isCompatibilityLoaded;
|
---|
[12332] | 77 | }
|
---|
[13184] | 78 | [Obsolete("The constructor of GBTModel should not be used directly anymore (use GBTModelSurrogate instead)")]
|
---|
[12332] | 79 | public GradientBoostedTreesModel(IEnumerable<IRegressionModel> models, IEnumerable<double> weights)
|
---|
[12372] | 80 | : base("Gradient boosted tree model", string.Empty) {
|
---|
[12332] | 81 | this.models = new List<IRegressionModel>(models);
|
---|
| 82 | this.weights = new List<double>(weights);
|
---|
| 83 |
|
---|
| 84 | if (this.models.Count != this.weights.Count) throw new ArgumentException();
|
---|
| 85 | }
|
---|
| 86 |
|
---|
| 87 | public override IDeepCloneable Clone(Cloner cloner) {
|
---|
| 88 | return new GradientBoostedTreesModel(this, cloner);
|
---|
| 89 | }
|
---|
| 90 |
|
---|
[12589] | 91 | public IEnumerable<double> GetEstimatedValues(IDataset dataset, IEnumerable<int> rows) {
|
---|
[12590] | 92 | // allocate target array go over all models and add up weighted estimation for each row
|
---|
[12660] | 93 | if (!rows.Any()) return Enumerable.Empty<double>(); // return immediately if rows is empty. This prevents multiple iteration over lazy rows enumerable.
|
---|
[13184] | 94 | // (which essentially looks up indexes in a dictionary)
|
---|
[12590] | 95 | var res = new double[rows.Count()];
|
---|
| 96 | for (int i = 0; i < models.Count; i++) {
|
---|
| 97 | var w = weights[i];
|
---|
| 98 | var m = models[i];
|
---|
| 99 | int r = 0;
|
---|
| 100 | foreach (var est in m.GetEstimatedValues(dataset, rows)) {
|
---|
| 101 | res[r++] += w * est;
|
---|
| 102 | }
|
---|
[12332] | 103 | }
|
---|
[12590] | 104 | return res;
|
---|
[12332] | 105 | }
|
---|
| 106 |
|
---|
| 107 | public IRegressionSolution CreateRegressionSolution(IRegressionProblemData problemData) {
|
---|
| 108 | return new RegressionSolution(this, (IRegressionProblemData)problemData.Clone());
|
---|
| 109 | }
|
---|
| 110 | }
|
---|
| 111 | }
|
---|