source: branches/CEDMA-Refactoring-Ticket419/HeuristicLab.CEDMA.Server/Dispatcher.cs @ 1181

#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 class Dispatcher {
    private List<Execution> dispatchQueue;
    public IList<string> DispatchQueue {
      get { return dispatchQueue.Select(t => "StandardGP").ToList(); }
    }

    private IStore store;

    public Dispatcher(IStore store) {
      this.store = store;
      this.dispatchQueue = new List<Execution>();
    }

    private void FillDispatchQueue() {
      // 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)
      };

      var dataSetBindings = store.Query("?DataSet <"+Ontology.PredicateInstanceOf.Uri+"> <"+Ontology.TypeDataSet.Uri+"> .");

      // no datasets => do nothing
      if (dataSetBindings.Count() == 0) return;

      // assume last dataset is the most interesting one
      // find and select all results for this dataset
      var dataSetEntity = (Entity)dataSetBindings.Last().Get("DataSet");
      var targetVar = new HeuristicLab.CEDMA.DB.Interfaces.Variable("TargetVariable");
      var modelVar = new HeuristicLab.CEDMA.DB.Interfaces.Variable("Model");
      var modelMAPE = new HeuristicLab.CEDMA.DB.Interfaces.Variable("ModelMAPE");

      var query = "<" + dataSetEntity.Uri + "> <" + Ontology.PredicateHasModel.Uri + "> ?Model ." + Environment.NewLine +
        "?Model <" + Ontology.TargetVariable.Uri + "> ?TargetVariable ." + Environment.NewLine +
        "?Model <" + Ontology.ValidationMeanAbsolutePercentageError.Uri + "> ?ModelMAPE .";



      var bindings = store.Query(query);
      DataSet dataSet = new DataSet(store, dataSetEntity);
      double[] utilization = new double[dataSet.Problem.AllowedTargetVariables.Count];
      int i = 0;
      int totalN = bindings.Count();
      foreach (int targetVariable in dataSet.Problem.AllowedTargetVariables) {
        var targetVarBindings = bindings.Where(x => (int)((Literal)x.Get("TargetVariable")).Value == targetVariable);
        if (targetVarBindings.Count() == 0) {
          utilization[i++] = double.PositiveInfinity;
        } else {
          double averageMape = targetVarBindings.Average(x => (double)((Literal)x.Get("ModelMAPE")).Value);
          double n = targetVarBindings.Count();
          utilization[i++] = -averageMape + Math.Sqrt(Math.Log(totalN) / n) * 0.1;
        }
      }
      int[] idx = Enumerable.Range(0, utilization.Length).ToArray();
      Array.Sort(utilization, idx);
      int nConfigurations = utilization.Length;
      for (int j = nConfigurations - 1; j > nConfigurations * 0.8; j--) {
        int targetVariable = dataSet.Problem.AllowedTargetVariables[idx[j]];
        IEngine engine = CreateEngine(dataSet.Problem, targetVariable);
        if (engine != null) {
          QueueJob(new Execution(dataSetEntity, engine, targetVariable));
        }
      }
    }

    private void QueueJob(Execution execution) {
      dispatchQueue.Add(execution);
    }

    public Execution GetNextJob() {
      if (dispatchQueue.Count == 0) {
        FillDispatchQueue();
      }
      if (dispatchQueue.Count > 0) {
        Execution next = dispatchQueue[0];
        dispatchQueue.RemoveAt(0);
        return next;
      } else
        return null;
    }

    internal void Start() {
      FillDispatchQueue();
    }

    private IEngine CreateEngine(Problem problem, int targetVariable) {
      switch (problem.LearningTask) {
        case LearningTask.Classification: return null;
        case LearningTask.Regression: {
            return CreateStandardGp(problem, targetVariable).Engine;
          }
        case LearningTask.TimeSeries: return null;
        case LearningTask.Clustering: return null;
        default: return null;
      }
    }

    private StandardGP CreateStandardGp(Problem problem, int targetVariable) {
      ProblemInjector probInjector = new ProblemInjector(problem);
      probInjector.TargetVariable = targetVariable;
      StandardGP sgp = new StandardGP();
      sgp.SetSeedRandomly = true;
      sgp.MaxGenerations = 300;
      sgp.PopulationSize = 10000;
      sgp.Elites = 1;
      sgp.ProblemInjector = probInjector;
      return sgp;
    }
  }
}