source: branches/LearningClassifierSystems/HeuristicLab.Encodings.DecisionList/3.3/Evaluator/MDLCalculator.cs @ 9334

#region License Information
/* HeuristicLab
 * Copyright (C) 2002-2012 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
 *
 * This file is part of HeuristicLab.
 *
 * HeuristicLab is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 *
 * HeuristicLab is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with HeuristicLab. If not, see <http://www.gnu.org/licenses/>.
 */
#endregion

using System.Collections.Generic;
using System.Linq;
using HeuristicLab.Common;
using HeuristicLab.Core;
using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;

namespace HeuristicLab.Encodings.DecisionList {
  [StorableClass]
  [Item("MDLCalculator", "")]
  public class MDLCalculator : Item {

    [Storable]
    private int activationIteration;

    [Storable]
    private bool activated;
    [Storable]
    private bool fixedWeight;

    [Storable]
    private double theoryWeight;

    [Storable]
    private double initialTheoryLengthRatio;
    [Storable]
    private double weightRelaxFactor;

    [Storable]
    private IList<double> accuracyStatistic;
    [Storable]
    private int iterationsSinceBest;
    [Storable]
    private double bestFitness;

    [StorableConstructor]
    protected MDLCalculator(bool deserializing) : base(deserializing) { }
    protected MDLCalculator(MDLCalculator original, Cloner cloner)
      : base(original, cloner) {
    }
    public MDLCalculator(int activationIteration, double initialTheoryLengthRatio, double weightRelaxFactor)
      : base() {
      this.activationIteration = activationIteration;
      this.initialTheoryLengthRatio = initialTheoryLengthRatio;
      this.weightRelaxFactor = weightRelaxFactor;
      activated = false;
      fixedWeight = false;
      accuracyStatistic = new List<double>();
      iterationsSinceBest = 0;
    }
    public override IDeepCloneable Clone(Cloner cloner) {
      return new MDLCalculator(this, cloner);
    }

    public void StartNewIteration(DecisionListSolution bestSolution, int currentIteration) {
      if (currentIteration == activationIteration) {
        activated = true;
        double error = bestSolution.TrainingExceptionsLength;
        double tl = (bestSolution.TrainingTheoryLength * bestSolution.ProblemData.Classes) / bestSolution.TrainingNumberOfAliveRules;
        theoryWeight = (initialTheoryLengthRatio / (1.0 - initialTheoryLengthRatio)) * (error / tl);
        iterationsSinceBest = 0;
      }

      if (activated && !fixedWeight &&
        GetLastIterationsAccuracyAverage(10) == 1.0) {
        fixedWeight = true;
      }

      if (activated && !fixedWeight) {
        if (bestSolution.TrainingAccuracy != 1.0) {
          if (iterationsSinceBest == 10) {
            theoryWeight *= weightRelaxFactor;
            iterationsSinceBest = 0;
          }
        }
      }

      if (currentIteration != 0) {
        UpdateStatistic(bestSolution.TrainingAccuracy);
      }
    }

    private void UpdateStatistic(double accuracy) {
      if (iterationsSinceBest == 0) {
        bestFitness = accuracy;
        iterationsSinceBest++;
      } else if (accuracy > bestFitness) {
        bestFitness = accuracy;
        iterationsSinceBest = 1;
      } else {
        iterationsSinceBest++;
      }
    }

    private double GetLastIterationsAccuracyAverage(int iterations) {
      int startAt = accuracyStatistic.Count - iterations;
      startAt = startAt > 0 ? startAt : 0;
      return accuracyStatistic.Skip(startAt).Sum() / (accuracyStatistic.Count - startAt);
    }

    public double CalculateFitness(DecisionListSolution dls) {
      double fitness = 0;
      if (activated) {
        fitness = dls.TrainingTheoryLength * theoryWeight;
      }
      fitness += 105.0 - dls.TrainingAccuracy * 100.0;
      return fitness;
    }
  }
}