Free cookie consent management tool by TermsFeed Policy Generator

source: branches/RemoveBackwardsCompatibility/HeuristicLab.Problems.DataAnalysis.Symbolic.Regression/3.4/SingleObjective/Evaluators/SymbolicRegressionLogResidualEvaluator.cs @ 13346

Last change on this file since 13346 was 12977, checked in by bburlacu, 9 years ago

#2480: Reverted changes to the evaluators as they impact performance and memory usage.

File size: 5.2 KB
Line 
1#region License Information
2/* HeuristicLab
3 * Copyright (C) 2002-2015 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.Core;
27using HeuristicLab.Data;
28using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
29using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
30
31namespace HeuristicLab.Problems.DataAnalysis.Symbolic.Regression {
32  [Item("Log Residual Evaluator", "Evaluator for symbolic regression models that calculates the mean of logarithmic absolute residuals avg(log( 1 + abs(y' - y)))" +
33                                  "This evaluator does not perform linear scaling!" +
34                                  "This evaluator can be useful if the modeled function contains discontinuities (e.g. 1/x). " +
35                                  "For some data sets (e.g. Korns benchmark instances containing inverses of near zero values) the squared error or absolute " +
36                                  "error put too much emphasis on modeling the outlier values. Using log-residuals instead has the " +
37                                  "effect that smaller residuals have a stronger impact on the total quality compared to the large residuals." +
38                                  "This effects GP convergence because functional fragments which are necessary to explain small variations are also more likely" +
39                                  " to stay in the population. This is useful even when the actual objective function is mean of squared errors.")]
40  [StorableClass]
41  public class SymbolicRegressionLogResidualEvaluator : SymbolicRegressionSingleObjectiveEvaluator {
42    [StorableConstructor]
43    protected SymbolicRegressionLogResidualEvaluator(bool deserializing) : base(deserializing) { }
44    protected SymbolicRegressionLogResidualEvaluator(SymbolicRegressionLogResidualEvaluator original, Cloner cloner)
45      : base(original, cloner) {
46    }
47    public override IDeepCloneable Clone(Cloner cloner) {
48      return new SymbolicRegressionLogResidualEvaluator(this, cloner);
49    }
50
51    public SymbolicRegressionLogResidualEvaluator() : base() { }
52
53    public override bool Maximization { get { return false; } }
54
55    public override IOperation InstrumentedApply() {
56      var solution = SymbolicExpressionTreeParameter.ActualValue;
57      IEnumerable<int> rows = GenerateRowsToEvaluate();
58
59      double quality = Calculate(SymbolicDataAnalysisTreeInterpreterParameter.ActualValue, solution, EstimationLimitsParameter.ActualValue.Lower, EstimationLimitsParameter.ActualValue.Upper, ProblemDataParameter.ActualValue, rows);
60      QualityParameter.ActualValue = new DoubleValue(quality);
61
62      return base.InstrumentedApply();
63    }
64
65    public static double Calculate(ISymbolicDataAnalysisExpressionTreeInterpreter interpreter, ISymbolicExpressionTree solution, double lowerEstimationLimit, double upperEstimationLimit, IRegressionProblemData problemData, IEnumerable<int> rows) {
66      IEnumerable<double> estimatedValues = interpreter.GetSymbolicExpressionTreeValues(solution, problemData.Dataset, rows);
67      IEnumerable<double> targetValues = problemData.Dataset.GetDoubleValues(problemData.TargetVariable, rows);
68      IEnumerable<double> boundedEstimatedValues = estimatedValues.LimitToRange(lowerEstimationLimit, upperEstimationLimit);
69
70      var logRes = boundedEstimatedValues.Zip(targetValues, (e, t) => Math.Log(1.0 + Math.Abs(e - t)));
71
72      OnlineCalculatorError errorState;
73      OnlineCalculatorError varErrorState;
74      double mlr;
75      double variance;
76      OnlineMeanAndVarianceCalculator.Calculate(logRes, out mlr, out variance, out errorState, out varErrorState);
77      if (errorState != OnlineCalculatorError.None) return double.NaN;
78      return mlr;
79    }
80
81    public override double Evaluate(IExecutionContext context, ISymbolicExpressionTree tree, IRegressionProblemData problemData, IEnumerable<int> rows) {
82      SymbolicDataAnalysisTreeInterpreterParameter.ExecutionContext = context;
83      EstimationLimitsParameter.ExecutionContext = context;
84
85      double mlr = Calculate(SymbolicDataAnalysisTreeInterpreterParameter.ActualValue, tree, EstimationLimitsParameter.ActualValue.Lower, EstimationLimitsParameter.ActualValue.Upper, problemData, rows);
86
87      SymbolicDataAnalysisTreeInterpreterParameter.ExecutionContext = null;
88      EstimationLimitsParameter.ExecutionContext = null;
89
90      return mlr;
91    }
92  }
93}
Note: See TracBrowser for help on using the repository browser.