[17154] | 1 | #region License Information
|
---|
| 2 | /* HeuristicLab
|
---|
[17180] | 3 | * Copyright (C) Heuristic and Evolutionary Algorithms Laboratory (HEAL)
|
---|
[17154] | 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 |
|
---|
| 22 | using System;
|
---|
| 23 | using System.Collections.Generic;
|
---|
| 24 | using System.Linq;
|
---|
| 25 | using HEAL.Attic;
|
---|
| 26 | using HeuristicLab.Common;
|
---|
| 27 | using HeuristicLab.Core;
|
---|
| 28 | using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
|
---|
| 29 | using HeuristicLab.Problems.DataAnalysis;
|
---|
| 30 | using HeuristicLab.Problems.DataAnalysis.Symbolic;
|
---|
| 31 |
|
---|
| 32 | namespace HeuristicLab.Algorithms.DataAnalysis {
|
---|
[17931] | 33 | [StorableType("55412E08-DAD4-4C2E-9181-C142E7EA9474")]
|
---|
[17154] | 34 | [Item("RandomForestModelFull", "Represents a random forest for regression and classification.")]
|
---|
| 35 | public sealed class RandomForestModelFull : ClassificationModel, IRandomForestModel {
|
---|
| 36 |
|
---|
| 37 | public override IEnumerable<string> VariablesUsedForPrediction {
|
---|
| 38 | get { return inputVariables; }
|
---|
| 39 | }
|
---|
| 40 |
|
---|
| 41 | [Storable]
|
---|
| 42 | private double[] classValues;
|
---|
| 43 |
|
---|
[17931] | 44 | public int NumClasses => classValues == null ? 0 : classValues.Length;
|
---|
| 45 |
|
---|
[17154] | 46 | [Storable]
|
---|
| 47 | private string[] inputVariables;
|
---|
| 48 |
|
---|
[17931] | 49 | [Storable]
|
---|
[17154] | 50 | public int NumberOfTrees {
|
---|
[17931] | 51 | get; private set;
|
---|
[17154] | 52 | }
|
---|
| 53 |
|
---|
| 54 | // not persisted
|
---|
| 55 | private alglib.decisionforest randomForest;
|
---|
| 56 |
|
---|
| 57 | [Storable]
|
---|
[17931] | 58 | private string RandomForestSerialized {
|
---|
| 59 | get { alglib.dfserialize(randomForest, out var serialized); return serialized; }
|
---|
| 60 | set { if (value != null) alglib.dfunserialize(value, out randomForest); }
|
---|
[17154] | 61 | }
|
---|
| 62 |
|
---|
| 63 | [StorableConstructor]
|
---|
[17931] | 64 | private RandomForestModelFull(StorableConstructorFlag _) : base(_) { }
|
---|
[17154] | 65 |
|
---|
| 66 | private RandomForestModelFull(RandomForestModelFull original, Cloner cloner) : base(original, cloner) {
|
---|
[17931] | 67 | if (original.randomForest != null)
|
---|
| 68 | randomForest = (alglib.decisionforest)original.randomForest.make_copy();
|
---|
| 69 | NumberOfTrees = original.NumberOfTrees;
|
---|
[17154] | 70 |
|
---|
| 71 | // following fields are immutable so we don't need to clone them
|
---|
| 72 | inputVariables = original.inputVariables;
|
---|
| 73 | classValues = original.classValues;
|
---|
| 74 | }
|
---|
| 75 | public override IDeepCloneable Clone(Cloner cloner) {
|
---|
| 76 | return new RandomForestModelFull(this, cloner);
|
---|
| 77 | }
|
---|
| 78 |
|
---|
[17931] | 79 | public RandomForestModelFull(alglib.decisionforest decisionForest, int nTrees, string targetVariable, IEnumerable<string> inputVariables, IEnumerable<double> classValues = null) : base(targetVariable) {
|
---|
[17154] | 80 | this.name = ItemName;
|
---|
| 81 | this.description = ItemDescription;
|
---|
| 82 |
|
---|
[17931] | 83 | randomForest = (alglib.decisionforest)decisionForest.make_copy();
|
---|
| 84 | NumberOfTrees = nTrees;
|
---|
[17154] | 85 |
|
---|
| 86 | this.inputVariables = inputVariables.ToArray();
|
---|
| 87 |
|
---|
| 88 | //classValues are only use for classification models
|
---|
| 89 | if (classValues == null) this.classValues = new double[0];
|
---|
| 90 | else this.classValues = classValues.ToArray();
|
---|
| 91 | }
|
---|
| 92 |
|
---|
| 93 |
|
---|
| 94 | public IRegressionSolution CreateRegressionSolution(IRegressionProblemData problemData) {
|
---|
| 95 | return new RandomForestRegressionSolution(this, new RegressionProblemData(problemData));
|
---|
| 96 | }
|
---|
| 97 | public override IClassificationSolution CreateClassificationSolution(IClassificationProblemData problemData) {
|
---|
| 98 | return new RandomForestClassificationSolution(this, new ClassificationProblemData(problemData));
|
---|
| 99 | }
|
---|
| 100 |
|
---|
| 101 | public bool IsProblemDataCompatible(IRegressionProblemData problemData, out string errorMessage) {
|
---|
| 102 | return RegressionModel.IsProblemDataCompatible(this, problemData, out errorMessage);
|
---|
| 103 | }
|
---|
| 104 |
|
---|
| 105 | public override bool IsProblemDataCompatible(IDataAnalysisProblemData problemData, out string errorMessage) {
|
---|
| 106 | if (problemData == null) throw new ArgumentNullException("problemData", "The provided problemData is null.");
|
---|
| 107 |
|
---|
| 108 | var regressionProblemData = problemData as IRegressionProblemData;
|
---|
| 109 | if (regressionProblemData != null)
|
---|
| 110 | return IsProblemDataCompatible(regressionProblemData, out errorMessage);
|
---|
| 111 |
|
---|
| 112 | var classificationProblemData = problemData as IClassificationProblemData;
|
---|
| 113 | if (classificationProblemData != null)
|
---|
| 114 | return IsProblemDataCompatible(classificationProblemData, out errorMessage);
|
---|
| 115 |
|
---|
| 116 | throw new ArgumentException("The problem data is not compatible with this random forest. Instead a " + problemData.GetType().GetPrettyName() + " was provided.", "problemData");
|
---|
| 117 | }
|
---|
| 118 |
|
---|
| 119 | public IEnumerable<double> GetEstimatedValues(IDataset dataset, IEnumerable<int> rows) {
|
---|
| 120 | double[,] inputData = dataset.ToArray(inputVariables, rows);
|
---|
| 121 | RandomForestUtil.AssertInputMatrix(inputData);
|
---|
| 122 |
|
---|
| 123 | int n = inputData.GetLength(0);
|
---|
| 124 | int columns = inputData.GetLength(1);
|
---|
| 125 | double[] x = new double[columns];
|
---|
| 126 | double[] y = new double[1];
|
---|
| 127 |
|
---|
[17931] | 128 | alglib.dfcreatebuffer(randomForest, out var buf);
|
---|
[17154] | 129 | for (int row = 0; row < n; row++) {
|
---|
| 130 | for (int column = 0; column < columns; column++) {
|
---|
| 131 | x[column] = inputData[row, column];
|
---|
| 132 | }
|
---|
[17931] | 133 | alglib.dftsprocess(randomForest, buf, x, ref y); // thread-safe process (as long as separate buffers are used)
|
---|
[17154] | 134 | yield return y[0];
|
---|
| 135 | }
|
---|
| 136 | }
|
---|
| 137 | public IEnumerable<double> GetEstimatedVariances(IDataset dataset, IEnumerable<int> rows) {
|
---|
| 138 | double[,] inputData = dataset.ToArray(inputVariables, rows);
|
---|
| 139 | RandomForestUtil.AssertInputMatrix(inputData);
|
---|
| 140 |
|
---|
| 141 | int n = inputData.GetLength(0);
|
---|
| 142 | int columns = inputData.GetLength(1);
|
---|
| 143 | double[] x = new double[columns];
|
---|
| 144 | double[] ys = new double[this.randomForest.innerobj.ntrees];
|
---|
| 145 |
|
---|
| 146 | for (int row = 0; row < n; row++) {
|
---|
| 147 | for (int column = 0; column < columns; column++) {
|
---|
| 148 | x[column] = inputData[row, column];
|
---|
| 149 | }
|
---|
[17931] | 150 | lock (randomForest)
|
---|
| 151 | alglib.dforest.dfprocessraw(randomForest.innerobj, x, ref ys, null);
|
---|
[17154] | 152 | yield return ys.VariancePop();
|
---|
| 153 | }
|
---|
| 154 | }
|
---|
| 155 |
|
---|
| 156 | public override IEnumerable<double> GetEstimatedClassValues(IDataset dataset, IEnumerable<int> rows) {
|
---|
| 157 | double[,] inputData = dataset.ToArray(inputVariables, rows);
|
---|
| 158 | RandomForestUtil.AssertInputMatrix(inputData);
|
---|
| 159 |
|
---|
| 160 | int n = inputData.GetLength(0);
|
---|
| 161 | int columns = inputData.GetLength(1);
|
---|
| 162 | double[] x = new double[columns];
|
---|
[17931] | 163 | double[] y = new double[NumClasses];
|
---|
[17154] | 164 |
|
---|
[17931] | 165 | alglib.dfcreatebuffer(randomForest, out var buf);
|
---|
[17154] | 166 | for (int row = 0; row < n; row++) {
|
---|
| 167 | for (int column = 0; column < columns; column++) {
|
---|
| 168 | x[column] = inputData[row, column];
|
---|
| 169 | }
|
---|
[17931] | 170 | alglib.dftsprocess(randomForest, buf, x, ref y);
|
---|
[17154] | 171 | // find class for with the largest probability value
|
---|
| 172 | int maxProbClassIndex = 0;
|
---|
| 173 | double maxProb = y[0];
|
---|
| 174 | for (int i = 1; i < y.Length; i++) {
|
---|
| 175 | if (maxProb < y[i]) {
|
---|
| 176 | maxProb = y[i];
|
---|
| 177 | maxProbClassIndex = i;
|
---|
| 178 | }
|
---|
| 179 | }
|
---|
| 180 | yield return classValues[maxProbClassIndex];
|
---|
| 181 | }
|
---|
| 182 | }
|
---|
| 183 |
|
---|
| 184 | public ISymbolicExpressionTree ExtractTree(int treeIdx) {
|
---|
| 185 | var rf = randomForest;
|
---|
| 186 | // hoping that the internal representation of alglib is stable
|
---|
| 187 |
|
---|
| 188 | // TREE FORMAT
|
---|
| 189 | // W[Offs] - size of sub-array (for the tree)
|
---|
| 190 | // node info:
|
---|
| 191 | // W[K+0] - variable number (-1 for leaf mode)
|
---|
| 192 | // W[K+1] - threshold (class/value for leaf node)
|
---|
| 193 | // W[K+2] - ">=" branch index (absent for leaf node)
|
---|
| 194 |
|
---|
| 195 | // skip irrelevant trees
|
---|
| 196 | int offset = 0;
|
---|
| 197 | for (int i = 0; i < treeIdx - 1; i++) {
|
---|
| 198 | offset = offset + (int)Math.Round(rf.innerobj.trees[offset]);
|
---|
| 199 | }
|
---|
| 200 |
|
---|
[18132] | 201 | var numSy = new Number();
|
---|
[17154] | 202 | var varCondSy = new VariableCondition() { IgnoreSlope = true };
|
---|
| 203 |
|
---|
[18132] | 204 | var node = CreateRegressionTreeRec(rf.innerobj.trees, offset, offset + 1, numSy, varCondSy);
|
---|
[17154] | 205 |
|
---|
| 206 | var startNode = new StartSymbol().CreateTreeNode();
|
---|
| 207 | startNode.AddSubtree(node);
|
---|
| 208 | var root = new ProgramRootSymbol().CreateTreeNode();
|
---|
| 209 | root.AddSubtree(startNode);
|
---|
| 210 | return new SymbolicExpressionTree(root);
|
---|
| 211 | }
|
---|
| 212 |
|
---|
[18132] | 213 | private ISymbolicExpressionTreeNode CreateRegressionTreeRec(double[] trees, int offset, int k, Number numSy, VariableCondition varCondSy) {
|
---|
[17154] | 214 |
|
---|
| 215 | // alglib source for evaluation of one tree (dfprocessinternal)
|
---|
| 216 | // offs = 0
|
---|
| 217 | //
|
---|
| 218 | // Set pointer to the root
|
---|
| 219 | //
|
---|
| 220 | // k = offs + 1;
|
---|
| 221 | //
|
---|
| 222 | // //
|
---|
| 223 | // // Navigate through the tree
|
---|
| 224 | // //
|
---|
| 225 | // while (true) {
|
---|
| 226 | // if ((double)(df.trees[k]) == (double)(-1)) {
|
---|
| 227 | // if (df.nclasses == 1) {
|
---|
| 228 | // y[0] = y[0] + df.trees[k + 1];
|
---|
| 229 | // } else {
|
---|
| 230 | // idx = (int)Math.Round(df.trees[k + 1]);
|
---|
| 231 | // y[idx] = y[idx] + 1;
|
---|
| 232 | // }
|
---|
| 233 | // break;
|
---|
| 234 | // }
|
---|
| 235 | // if ((double)(x[(int)Math.Round(df.trees[k])]) < (double)(df.trees[k + 1])) {
|
---|
| 236 | // k = k + innernodewidth;
|
---|
| 237 | // } else {
|
---|
| 238 | // k = offs + (int)Math.Round(df.trees[k + 2]);
|
---|
| 239 | // }
|
---|
| 240 | // }
|
---|
| 241 |
|
---|
[18132] | 242 | if (trees[k] == -1) {
|
---|
| 243 | var numNode = (NumberTreeNode)numSy.CreateTreeNode();
|
---|
| 244 | numNode.Value = trees[k + 1];
|
---|
| 245 | return numNode;
|
---|
[17154] | 246 | } else {
|
---|
| 247 | var condNode = (VariableConditionTreeNode)varCondSy.CreateTreeNode();
|
---|
| 248 | condNode.VariableName = inputVariables[(int)Math.Round(trees[k])];
|
---|
| 249 | condNode.Threshold = trees[k + 1];
|
---|
| 250 | condNode.Slope = double.PositiveInfinity;
|
---|
| 251 |
|
---|
[18132] | 252 | var left = CreateRegressionTreeRec(trees, offset, k + 3, numSy, varCondSy);
|
---|
| 253 | var right = CreateRegressionTreeRec(trees, offset, offset + (int)Math.Round(trees[k + 2]), numSy, varCondSy);
|
---|
[17154] | 254 |
|
---|
| 255 | condNode.AddSubtree(left); // not 100% correct because interpreter uses: if(x <= thres) left() else right() and RF uses if(x < thres) left() else right() (see above)
|
---|
| 256 | condNode.AddSubtree(right);
|
---|
| 257 | return condNode;
|
---|
| 258 | }
|
---|
| 259 | }
|
---|
| 260 | }
|
---|
| 261 | } |
---|