source: trunk/sources/HeuristicLab.CEDMA.Server/DispatcherBase.cs @ 1394

#region License Information
/* HeuristicLab
 * Copyright (C) 2002-2008 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;
using System.Collections.Generic;
using System.Text;
using System.Windows.Forms;
using HeuristicLab.PluginInfrastructure;
using System.Net;
using System.ServiceModel;
using HeuristicLab.CEDMA.DB.Interfaces;
using HeuristicLab.CEDMA.DB;
using System.ServiceModel.Description;
using System.Linq;
using HeuristicLab.CEDMA.Core;
using HeuristicLab.GP.StructureIdentification;
using HeuristicLab.Data;
using HeuristicLab.Core;

namespace HeuristicLab.CEDMA.Server {
  public abstract class DispatcherBase : IDispatcher {
    public enum ModelComplexity { Low, Medium, High };
    public enum Algorithm { StandardGpRegression, OffspringGpRegression, StandardGpClassification, OffspringGpClassification, StandardGpForecasting, OffspringGpForecasting };

    private IStore store;
    private ModelComplexity[] possibleComplexities = new ModelComplexity[] { ModelComplexity.Low, ModelComplexity.Medium, ModelComplexity.High };
    private Dictionary<LearningTask, Algorithm[]> possibleAlgorithms = new Dictionary<LearningTask, Algorithm[]>() {
      {LearningTask.Classification, new Algorithm[] { Algorithm.StandardGpClassification, Algorithm.OffspringGpClassification }},
      {LearningTask.Regression, new Algorithm[] { Algorithm.StandardGpRegression, Algorithm.OffspringGpRegression }},
      {LearningTask.TimeSeries, new Algorithm[] { Algorithm.StandardGpForecasting, Algorithm.OffspringGpForecasting }}
    };

    private static int MaxGenerations {
      get { return 3; }
    }

    private static int MaxEvaluatedSolutions {
      get { return 3000; }
    }

    public DispatcherBase(IStore store) {
      this.store = store;
    }

    public Execution GetNextJob() {
      // find and select a dataset
      var dataSetVar = new HeuristicLab.CEDMA.DB.Interfaces.Variable("DataSet");
      var dataSetQuery = new Statement[] {
        new Statement(dataSetVar, Ontology.PredicateInstanceOf, Ontology.TypeDataSet)
      };

      Entity[] datasets = store.Query("?DataSet <" + Ontology.PredicateInstanceOf.Uri + "> <" + Ontology.TypeDataSet.Uri + "> .")
        .Select(x => (Entity)x.Get("DataSet"))
        .ToArray();

      // no datasets => do nothing
      if (datasets.Length == 0) return null;

      Entity dataSetEntity = SelectDataSet(datasets);
      DataSet dataSet = new DataSet(store, dataSetEntity);

      int targetVariable = SelectTargetVariable(dataSet, dataSet.Problem.AllowedTargetVariables.ToArray());
      Algorithm selectedAlgorithm = SelectAlgorithm(dataSet, targetVariable, possibleAlgorithms[dataSet.Problem.LearningTask]);
      string targetVariableName = dataSet.Problem.GetVariableName(targetVariable);
      ModelComplexity selectedComplexity = SelectComplexity(dataSet, targetVariable, selectedAlgorithm, possibleComplexities);

      Execution exec = CreateExecution(dataSet.Problem, targetVariable, selectedAlgorithm, selectedComplexity);
      if (exec != null) {
        exec.DataSetEntity = dataSetEntity;
        exec.TargetVariable = targetVariableName;
      }
      return exec;
    }

    public abstract Entity SelectDataSet(Entity[] datasets);
    public abstract int SelectTargetVariable(DataSet dataSet, int[] targetVariables);
    public abstract Algorithm SelectAlgorithm(DataSet dataSet, int targetVariable, Algorithm[] possibleAlgorithms);
    public abstract ModelComplexity SelectComplexity(DataSet dataSet, int targetVariable, Algorithm algorithm, ModelComplexity[] possibleComplexities);

    private Execution CreateExecution(Problem problem, int targetVariable, Algorithm algorithm, ModelComplexity complexity) {
      switch (algorithm) {
        case Algorithm.StandardGpRegression: {
            var algo = new HeuristicLab.GP.StructureIdentification.StandardGP();
            SetComplexityParameters(algo, complexity);
            SetProblemParameters(algo, problem, targetVariable);
            algo.PopulationSize = 10000;
            algo.MaxGenerations = MaxGenerations;
            Execution exec = new Execution(algo.Engine);
            exec.Description = "StandardGP - Complexity: " + complexity;
            return exec;
          }
        case Algorithm.OffspringGpRegression: {
            var algo = new HeuristicLab.GP.StructureIdentification.OffspringSelectionGP();
            SetComplexityParameters(algo, complexity);
            SetProblemParameters(algo, problem, targetVariable);
            algo.MaxEvaluatedSolutions = MaxEvaluatedSolutions;
            Execution exec = new Execution(algo.Engine);
            exec.Description = "OffspringGP - Complexity: " + complexity;
            return exec;
          }
        case Algorithm.StandardGpClassification: {
            var algo = new HeuristicLab.GP.StructureIdentification.Classification.StandardGP();
            SetComplexityParameters(algo, complexity);
            SetProblemParameters(algo, problem, targetVariable);
            algo.PopulationSize = 10000;
            algo.MaxGenerations = MaxGenerations;
            Execution exec = new Execution(algo.Engine);
            exec.Description = "StandardGP - Complexity: " + complexity;
            return exec;
          }
        case Algorithm.OffspringGpClassification: {
            var algo = new HeuristicLab.GP.StructureIdentification.Classification.OffspringSelectionGP();
            SetComplexityParameters(algo, complexity);
            SetProblemParameters(algo, problem, targetVariable);
            algo.MaxEvaluatedSolutions = MaxEvaluatedSolutions;
            Execution exec = new Execution(algo.Engine);
            exec.Description = "OffspringGP - Complexity: " + complexity;
            return exec;
          }
        case Algorithm.StandardGpForecasting: {
            var algo = new HeuristicLab.GP.StructureIdentification.TimeSeries.StandardGP();
            SetComplexityParameters(algo, complexity);
            SetProblemParameters(algo, problem, targetVariable);
            algo.PopulationSize = 10000;
            algo.MaxGenerations = MaxGenerations;
            Execution exec = new Execution(algo.Engine);
            exec.Description = "StandardGP - Complexity: " + complexity;
            return exec;
          }
        case Algorithm.OffspringGpForecasting: {
            var algo = new HeuristicLab.GP.StructureIdentification.TimeSeries.OffspringSelectionGP();
            SetComplexityParameters(algo, complexity);
            SetProblemParameters(algo, problem, targetVariable);
            algo.MaxEvaluatedSolutions = MaxEvaluatedSolutions;
            Execution exec = new Execution(algo.Engine);
            exec.Description = "OffspringGP - Complexity: " + complexity;
            return exec;
          }
        default: {
            return null;
          }
      }
    }

    private void SetComplexityParameters(AlgorithmBase algo, ModelComplexity complexity) {
      switch (complexity) {
        case ModelComplexity.Low: {
            algo.MaxTreeHeight = 5;
            algo.MaxTreeSize = 20;
            break;
          }
        case ModelComplexity.Medium: {
            algo.MaxTreeHeight = 10;
            algo.MaxTreeSize = 100;
            break;
          }
        case ModelComplexity.High: {
            algo.MaxTreeHeight = 12;
            algo.MaxTreeSize = 200;
            break;
          }
      }
    }

    private void SetProblemParameters(AlgorithmBase algo, Problem problem, int targetVariable) {
      algo.ProblemInjector.GetVariable("Dataset").Value = problem.DataSet;
      algo.ProblemInjector.GetVariable("TargetVariable").GetValue<IntData>().Data = targetVariable;
      algo.ProblemInjector.GetVariable("TrainingSamplesStart").GetValue<IntData>().Data = problem.TrainingSamplesStart;
      algo.ProblemInjector.GetVariable("TrainingSamplesEnd").GetValue<IntData>().Data = problem.TrainingSamplesEnd;
      algo.ProblemInjector.GetVariable("ValidationSamplesStart").GetValue<IntData>().Data = problem.ValidationSamplesStart;
      algo.ProblemInjector.GetVariable("ValidationSamplesEnd").GetValue<IntData>().Data = problem.ValidationSamplesEnd;
      algo.ProblemInjector.GetVariable("TestSamplesStart").GetValue<IntData>().Data = problem.TestSamplesStart;
      algo.ProblemInjector.GetVariable("TestSamplesEnd").GetValue<IntData>().Data = problem.TestSamplesEnd;
      ItemList<IntData> allowedFeatures = algo.ProblemInjector.GetVariable("AllowedFeatures").GetValue<ItemList<IntData>>();
      foreach (int allowedFeature in problem.AllowedInputVariables) allowedFeatures.Add(new IntData(allowedFeature));

      if (problem.LearningTask == LearningTask.TimeSeries) {
        algo.ProblemInjector.GetVariable("Autoregressive").GetValue<BoolData>().Data = problem.AutoRegressive;
        algo.ProblemInjector.GetVariable("MinTimeOffset").GetValue<IntData>().Data = problem.MinTimeOffset;
        algo.ProblemInjector.GetVariable("MaxTimeOffset").GetValue<IntData>().Data = problem.MaxTimeOffset;
      } else if (problem.LearningTask == LearningTask.Classification) {
        ItemList<DoubleData> classValues = algo.ProblemInjector.GetVariable("TargetClassValues").GetValue<ItemList<DoubleData>>();
        foreach (double classValue in GetDifferentClassValues(problem.DataSet, targetVariable)) classValues.Add(new DoubleData(classValue));
      }
    }

    private IEnumerable<double> GetDifferentClassValues(HeuristicLab.DataAnalysis.Dataset dataset, int targetVariable) {
      return Enumerable.Range(0, dataset.Rows).Select(x => dataset.GetValue(x, targetVariable)).Distinct();
    }
  }
}