source: branches/2988_ModelsOfModels2/HeuristicLab.Algorithms.EMM/ModelSetPreporation.cs @ 18065

using HEAL.Attic;
using HeuristicLab.Algorithms.DataAnalysis;
using HeuristicLab.Common;
using HeuristicLab.Core;
using HeuristicLab.Data;
using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
using HeuristicLab.Optimization;
using HeuristicLab.Parameters;
using HeuristicLab.Problems.DataAnalysis;
using HeuristicLab.Problems.DataAnalysis.Symbolic;
using HeuristicLab.Random;
using System;
using System.Collections.Generic;
using System.IO;
using System.Linq;
using CancellationToken = System.Threading.CancellationToken;
using ExecutionContext = HeuristicLab.Core.ExecutionContext;

namespace HeuristicLab.Algorithms.EvolvmentModelsOfModels {
  [Item("ModelSetPreparation", "Model Set preparation algorithm.")]
  [Creatable(CreatableAttribute.Categories.Algorithms, Priority = 125)]
  [StorableType("3C5DF308-DB79-4ACD-894B-F795F081726B")]
  public class ModelSetPreparation : FixedDataAnalysisAlgorithm<ISymbolicDataAnalysisSingleObjectiveProblem> {
    #region data members
    [Storable]
    protected ExecutionContext executionContext;
    [Storable]
    protected ExecutionState previousExecutionState;
    [Storable]
    protected IEnumerable<ISymbolicExpressionTree> trees;
    [Storable]
    protected ExecutionState currentExecutionState;
    #endregion

    #region parameters
    private const string SeedParameterName = "Seed";
    private const string SetSeedRandomlyParameterName = "SetSeedRandomly";
    private const string RandomParameterName = "Random";
    private const string InputFileParameterName = "InputFile";
    private const string AlgorithmImplementationTypeParameterName = "AlgorithmImplementationType";
    private const string GoalParameterName = "Goal";
    private const string DistanceTypeParameterName = "DistanceType";
    private const string MapParameterName = "Map";

    public IFixedValueParameter<IntValue> SeedParameter {
      get { return (IFixedValueParameter<IntValue>)Parameters[SeedParameterName]; }
    }
    public IConstrainedValueParameter<StringValue> AlgorithmImplementationTypeParameter {
      get { return (IConstrainedValueParameter<StringValue>)Parameters[AlgorithmImplementationTypeParameterName]; }
    }
    public IConstrainedValueParameter<StringValue> GoalParameter {
      get { return (IConstrainedValueParameter<StringValue>)Parameters[GoalParameterName]; }
    }
    public IConstrainedValueParameter<StringValue> DistanceTypeParameter {
      get { return (IConstrainedValueParameter<StringValue>)Parameters[DistanceTypeParameterName]; }
    }
    public IConstrainedValueParameter<EMMMapBase<ISymbolicExpressionTree>> MapParameter {
      get { return (IConstrainedValueParameter<EMMMapBase<ISymbolicExpressionTree>>)Parameters[MapParameterName]; }
    }
    public IFixedValueParameter<StringValue> InputFileParameter {
      get { return (IFixedValueParameter<StringValue>)Parameters[InputFileParameterName]; }
    }
    public IFixedValueParameter<BoolValue> SetSeedRandomlyParameter {
      get { return (IFixedValueParameter<BoolValue>)Parameters[SetSeedRandomlyParameterName]; }
    }
    public IValueParameter<IRandom> RandomParameter {
      get { return (IValueParameter<IRandom>)Parameters[RandomParameterName]; }
    }
    #endregion

    #region parameter properties
    public int Seed {
      get { return SeedParameter.Value.Value; }
      set { SeedParameter.Value.Value = value; }
    }
    public StringValue AlgorithmImplemetationType {
      get { return AlgorithmImplementationTypeParameter.Value; }
      set { AlgorithmImplementationTypeParameter.Value.Value = value.Value; }
    }
    public StringValue Goal {
      get { return GoalParameter.Value; }
      set { GoalParameter.Value.Value = value.Value; }
    }
    public StringValue DistanceType {
      get { return DistanceTypeParameter.Value; }
      set { DistanceTypeParameter.Value.Value = value.Value; }
    }
    public EMMMapBase<ISymbolicExpressionTree> Map {
      get { return MapParameter.Value; }
      set { MapParameter.Value = value; }
    }
    public StringValue InputFile {
      get { return InputFileParameter.Value; }
      set { InputFileParameter.Value.Value = value.Value; }
    }
    public bool SetSeedRandomly {
      get { return SetSeedRandomlyParameter.Value.Value; }
      set { SetSeedRandomlyParameter.Value.Value = value; }
    }
    #endregion

    #region constructors
    public ModelSetPreparation() {

      Parameters.Add(new FixedValueParameter<IntValue>(SeedParameterName, "The random seed used to initialize the new pseudo random number generator.", new IntValue(0)));
      Parameters.Add(new FixedValueParameter<StringValue>(InputFileParameterName, "The file with set of models that will be.", new StringValue("input.txt")));
      Parameters.Add(new ConstrainedValueParameter<StringValue>(AlgorithmImplementationTypeParameterName, "The Type of possible algorithm implementation, choose one: OnlyMap, Full, Read."));
      Parameters.Add(new ConstrainedValueParameter<StringValue>(GoalParameterName, "The goal of algorithm implementation, choose one: ToSee, ToWork, Full."));
      Parameters.Add(new ConstrainedValueParameter<StringValue>(DistanceTypeParameterName, "The Type of possible distance calculator for case of only distance calculation."));
      Parameters.Add(new ConstrainedValueParameter<EMMMapBase<ISymbolicExpressionTree>>(MapParameterName, "The type of map creation algorithm. Use one from: IslandMap, NetworkMap."));
      Parameters.Add(new FixedValueParameter<BoolValue>(SetSeedRandomlyParameterName, "True if the random seed should be set to a random value, otherwise false.", new BoolValue(true)));

      Parameters.Add(new ValueParameter<IRandom>(RandomParameterName, new MersenneTwister()));

      //begin hack ...
      InputFile.ValueChanged += InputFile_ValueChanged;
      InfixExpressionParser parser = new InfixExpressionParser();
      trees = File.ReadAllLines(InputFileParameter.Value.Value).Select(parser.Parse).ToArray();
      // end hack

      ProblemChanged += ModelSetPreporation_ProblemChanged;
      MapParameterUpdate();

    }

    // also hack !!!!!!!!!!!!!!!!!!!!!!!!!
    private void InputFile_ValueChanged(object sender, EventArgs e) {
      InfixExpressionParser parser = new InfixExpressionParser();
      trees = File.ReadAllLines(InputFileParameter.Value.Value).Select(parser.Parse);
    }
    // remove again !!!!!!!!!!!!!!!!!!!!!! 

    private void ModelSetPreporation_ProblemChanged(object sender, EventArgs e) {
      if (Problem != null) {
        Problem.SymbolicExpressionTreeInterpreter = new SymbolicDataAnalysisExpressionTreeBatchInterpreter();
      }
    }
    protected void MapParameterUpdate() {

      var mapTypes = new EMMMapBase<ISymbolicExpressionTree>[]
      {
        new EMMZeroMap (),
        new EMMIslandMap(),
        new EMMNetworkMap(),
        new EMMDisatanceMap(),
        new EMMRankMap(),
        new EMMSucsessMap ()
      };
      foreach (var t in mapTypes) {
        MapParameter.ValidValues.Add(t);
      }
      var algorithmType = new StringValue[]
        {
          new StringValue ("DistanceCalculation"),
          new StringValue ("OnlyMap"),
          new StringValue ("Statistic")
        };
      foreach (var t in algorithmType) {
        AlgorithmImplementationTypeParameter.ValidValues.Add(t);
      }
      var goal = new StringValue[]
  {
          new StringValue ("ToWork"),
          new StringValue ("ToSee"),
          new StringValue ("Full")
  };
      foreach (var t in goal) {
        GoalParameter.ValidValues.Add(t);
      }
      var distanceType = new StringValue[]
        {
          new StringValue("MSE"),
          new StringValue("PearsonsRSquared"),
          new StringValue ("Covariance"),
          new StringValue ("MaxAbsoluteError"),
          new StringValue ("MeanAbsoluteError"),
          new StringValue ("Symbolic")
        };
      foreach (var t in distanceType) {
        DistanceTypeParameter.ValidValues.Add(t);
      }
    }

    protected ModelSetPreparation(ModelSetPreparation original, Cloner cloner) : base(original, cloner) {

      previousExecutionState = original.previousExecutionState;
      if (original.executionContext != null) {
        executionContext = cloner.Clone(original.executionContext);
      }
      // hack
      trees = original.trees.Select(x => cloner.Clone(x)).ToArray();
    }

    [StorableConstructor]
    protected ModelSetPreparation(StorableConstructorFlag _) : base(_) { }
    public override IDeepCloneable Clone(Cloner cloner) {
      return new ModelSetPreparation(this, cloner);
    }
    #endregion

    #region technical stuff
    public override void Prepare() {
      base.Prepare();
    }

    protected override void Initialize(CancellationToken cancellationToken) {
      base.Initialize(cancellationToken);
    }

    public override bool SupportsPause => true;

    // implements random number generation from https://en.wikipedia.org/wiki/Dirichlet_distribution#Random_number_generation

    #region operator wiring and events
    private void ParameterizeStochasticOperator(IOperator op) {
      IStochasticOperator stochasticOp = op as IStochasticOperator;
      if (stochasticOp != null) {
        stochasticOp.RandomParameter.ActualName = "Random";
        stochasticOp.RandomParameter.Hidden = true;
      }
    }
    protected void ExecuteOperation(ExecutionContext executionContext, CancellationToken cancellationToken, IOperation operation) {
      Stack<IOperation> executionStack = new Stack<IOperation>();
      executionStack.Push(operation);
      while (executionStack.Count > 0) {
        cancellationToken.ThrowIfCancellationRequested();
        IOperation next = executionStack.Pop();
        if (next is OperationCollection coll) {
          for (int i = coll.Count - 1; i >= 0; i--)
            if (coll[i] != null) executionStack.Push(coll[i]);
        } else if (next is IAtomicOperation op) {
          next = op.Operator.Execute((IExecutionContext)op, cancellationToken);
          if (next != null) executionStack.Push(next);
        }
      }
    }

    protected override void OnProblemChanged() {
      base.OnProblemChanged();
    }

    protected override void OnExecutionStateChanged() {
      previousExecutionState = currentExecutionState;
      currentExecutionState = ExecutionState;
      base.OnExecutionStateChanged();
    }

    protected override void OnStopped() {
      if (executionContext != null) {
        if (executionContext.Scope != null) {
          if (executionContext.Scope.SubScopes != null) {
            executionContext.Scope.SubScopes.Clear();
          }
        }
      }
      base.OnStopped();
    }
    #endregion
    #endregion

    #region algorithm implementation
    protected override void Run(CancellationToken cancellationToken) {
      Map.DistanceParametr = DistanceType.Value;
      //distance calculation or reading that should be done in any cases
      string fileNameForWatch = "DistanceMatrix_Watch" + DistanceType + ".txt";
      string fileName = "DistanceMatrix_" + DistanceType + ".txt";
      double[,] totalDistance;
      if (AlgorithmImplemetationType.Value == "DistanceCalculation") {
        totalDistance = TotalDistanceMatrixCalculation(RandomParameter.Value, Problem, trees.ToList(), DistanceType.Value);
        if (Goal.Value != "ToWork") {
          FileComuncations.DoubleMatrixPrint(fileNameForWatch, totalDistance, trees.Count());
        }
        if (Goal.Value != "ToSee") {
          FileComuncations.DoubleMatrixSerialize(fileName, totalDistance);
        }
      } else {
        totalDistance = FileComuncations.DoubleMatrixDeserialize(fileName);
        // totalDistance = FileComuncations.DoubleMatrixFromFileRead(fileName, trees.Count());
      }

      if (AlgorithmImplemetationType.Value == "Statistic") {
        var statistic = new int[trees.Count(), trees.Count()];
        for (int i = 0; i < trees.Count(); i++) {
          for (int j = 0; j < trees.Count(); j++)
            statistic[i, j] = 0;
        }
        var maps = new List<List<List<int>>>();
        int repetitionNumber = 10;
        Map.MapCreationPrepare(trees);
        for (int i = 0; i < repetitionNumber; i++) {
          Map.CreateMap(RandomParameter.Value, totalDistance);
          maps.Add(Map.Map);
          CheckClusters(statistic);
          Map.Map.Clear();
        }
      } else { // Simple map creation case
        Map.MapCreationPrepare(trees);
        Map.CreateMap(RandomParameter.Value, totalDistance);
        Map.WriteMapToTxtFile(RandomParameter.Value);// This should be deactivated in case of using HIVE. HIVE can not work with it.
      }
    }
    protected void CheckClusters(int[,] info) {
      // ToDo: It should be realized for statistics collection
    }
    #region distance manipulation

    public static double[,] DistanceMatrixCalculation(List<ISymbolicExpressionTree> trees, string distanceType, ISymbolicDataAnalysisSingleObjectiveProblem Problem) {
      var problemData = (IRegressionProblemData)Problem.ProblemData;
      var dataset = problemData.Dataset;
      var rows = problemData.TrainingIndices;
      var interpreter = Problem.SymbolicExpressionTreeInterpreter;
      string[] toWrite = new string[trees.Count()];
      int i = 0;
      var treeValues = new List<List<double>>();
      if (distanceType != "Symbolic") {
        foreach (var tree in trees) {
          treeValues.Add(interpreter.GetSymbolicExpressionTreeValues(tree, dataset, rows).ToList());
        }
      }
      double[,] distances = new double[trees.Count, trees.Count];
      OnlineCalculatorError err;
      switch (distanceType) {
        case "MSE":
          for (i = 0; i < trees.Count - 1; i++) {
            for (int j = i + 1; j < trees.Count; j++) {
              distances[j, i] = distances[i, j] = OnlineNormalizedMeanSquaredErrorCalculator.Calculate(treeValues[i], treeValues[j], out err); ;
            }
          }
          break;
        case "PearsonsRSquared":
          for (i = 0; i < trees.Count - 1; i++) {
            for (int j = i + 1; j < trees.Count; j++) {
              distances[j, i] = distances[i, j] = OnlinePearsonsRCalculator.Calculate(treeValues[i], treeValues[j], out err); ;
            }
          }
          break;
        case "Covariance":
          for (i = 0; i < trees.Count - 1; i++) {
            for (int j = i + 1; j < trees.Count; j++) {
              distances[j, i] = distances[i, j] = OnlineCovarianceCalculator.Calculate(treeValues[i], treeValues[j], out err); ;
            }
          }
          break;
        case "MaxAbsoluteError":
          for (i = 0; i < trees.Count - 1; i++) {
            for (int j = i + 1; j < trees.Count; j++) {
              distances[j, i] = distances[i, j] = OnlineMaxAbsoluteErrorCalculator.Calculate(treeValues[i], treeValues[j], out err); ;
            }
          }
          break;
        case "MeanAbsoluteError":
          for (i = 0; i < trees.Count - 1; i++) {
            for (int j = i + 1; j < trees.Count; j++) {
              distances[j, i] = distances[i, j] = OnlineMeanAbsoluteErrorCalculator.Calculate(treeValues[i], treeValues[j], out err); ;
            }
          }
          break;
        case "Symbolic":
          distances = SymbolicExpressionTreeHash.ComputeSimilarityMatrix(trees, simplify: false, strict: true);
          for (i = 0; i < trees.Count - 1; i++) {
            for (int j = i + 1; j < trees.Count; j++) {
              distances[j, i] = distances[i, j] = 1 - distances[i, j];
            }
          }
          break;
      }

      return distances;
    }
    public static double[,] CalculateDistances(List<ISymbolicExpressionTree> treesSet) {
      double[,] distances;
      distances = SymbolicExpressionTreeHash.ComputeSimilarityMatrix(treesSet, simplify: false, strict: true);
      for (int i = 0; i < treesSet.Count - 1; i++) {
        for (int j = i + 1; j < treesSet.Count; j++) {
          distances[j, i] = distances[i, j] = 1 - distances[i, j];
        }
      }
      return distances;
    }
    public static double[,] TotalDistanceMatrixCalculation(IRandom random, ISymbolicDataAnalysisSingleObjectiveProblem problem, List<ISymbolicExpressionTree> treesSet, string distanceType) {
      var setSize = treesSet.Count();
      var totalDistance = new double[setSize, setSize];
      var treeSetTemp = new List<ISymbolicExpressionTree>();
      foreach (var tree in treesSet) {
        treeSetTemp.Add((ISymbolicExpressionTree)tree.Clone());
      }
      if (distanceType != "Symbolic") {

        int repitNumber = 10;
        totalDistance = new double[setSize, setSize];
        for (int i = 0; i < setSize; i++) {
          for (int j = 0; j < setSize; j++) {
            totalDistance[i, j] = 0;
          }
        }
        for (int i = 0; i < repitNumber; i++) {
          foreach (var tree in treeSetTemp) {
            HelpFunctions.SetLocalParametersForTree(random, 0.5, tree);
          }
          var distanceMatrix = DistanceMatrixCalculation(treeSetTemp, distanceType, problem);
          for (int t = 0; t < setSize; t++) {
            for (int j = 0; j < setSize; j++) {
              totalDistance[t, j] += Math.Abs(distanceMatrix[t, j]) / repitNumber;
            }
          }
        }
      } else {
        foreach (var tree in treeSetTemp) {
          HelpFunctions.SetLocalParametersForTree(random, 0.5, tree);
        }
        totalDistance = CalculateDistances(treeSetTemp);
      }
      return totalDistance;
    }
    #endregion 
    #endregion
  }
}