source: branches/HeuristicLab.EvolutionTracking/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Analyzers/BuildingBlockAnalyzers/SymbolicDataAnalysisPoly10Analyzer.cs @ 16130

#region License Information
/* HeuristicLab
 * Copyright (C) 2002-2014 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.Linq;
using HeuristicLab.Common;
using HeuristicLab.Core;
using HeuristicLab.Data;
using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
using HeuristicLab.Optimization;
using HeuristicLab.Parameters;
using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
using DataRow = HeuristicLab.Analysis.DataRow;
using DataTable = HeuristicLab.Analysis.DataTable;

namespace HeuristicLab.Problems.DataAnalysis.Symbolic.Analyzers.BuidingBlocks {
  [Item("Poly-10 building blocks analyzer", "An analyzer which attempts to identify parts of the Poly-10 formula")]
  [StorableClass]
  public class SymbolicDataAnalysisPoly10Analyzer : SymbolicDataAnalysisAnalyzer {
    private const string SymbolicDataAnalysisTreeInterpreterParameterName = "SymbolicExpressionTreeInterpreter";
    private const string ProblemDataParameterName = "ProblemData";
    private const string GenerationsParameterName = "Generations";
    private const string PhenotypicSimilarityThresholdParameterName = "PhenotypicSimilarityThreshold";
    private const string UpdateCounterParameterName = "UpdateCounter";
    private const string UpdateIntervalParameterName = "UpdateInterval";
    private const string BuildingBlocksFrequenciesTableName = "Building blocks frequencies";

    // store evaluations of building blocks for phenotypic matching
    private readonly Dictionary<string, List<double>> evaluationMap = new Dictionary<string, List<double>>();
    private readonly Dictionary<string, ISymbolicExpressionTreeNode> fragmentMap = new Dictionary<string, ISymbolicExpressionTreeNode>();
    private readonly Dictionary<string, string> prettyLabels = new Dictionary<string, string>();
    private readonly SymbolicExpressionImporter importer = new SymbolicExpressionImporter();

    #region Parameters
    public IValueParameter<DoubleValue> PhenotypicSimilarityThresholdParameter {
      get { return (IValueParameter<DoubleValue>)Parameters[PhenotypicSimilarityThresholdParameterName]; }
    }

    public ILookupParameter<ISymbolicDataAnalysisExpressionTreeInterpreter> SymbolicDataAnalysisTreeInterpreterParameter {
      get { return (ILookupParameter<ISymbolicDataAnalysisExpressionTreeInterpreter>)Parameters[SymbolicDataAnalysisTreeInterpreterParameterName]; }
    }

    public ILookupParameter<IDataAnalysisProblemData> ProblemDataParameter {
      get { return (ILookupParameter<IDataAnalysisProblemData>)Parameters[ProblemDataParameterName]; }
    }

    public ILookupParameter<IntValue> GenerationsParameter {
      get { return (ILookupParameter<IntValue>)Parameters[GenerationsParameterName]; }
    }

    public ValueParameter<IntValue> UpdateCounterParameter {
      get { return (ValueParameter<IntValue>)Parameters[UpdateCounterParameterName]; }
    }

    public ValueParameter<IntValue> UpdateIntervalParameter {
      get { return (ValueParameter<IntValue>)Parameters[UpdateIntervalParameterName]; }
    }
    #endregion

    #region Parameter properties
    public double PhenotypicSimilarityThreshold {
      get { return PhenotypicSimilarityThresholdParameter.Value.Value; }
      set { PhenotypicSimilarityThresholdParameter.Value.Value = value; }
    }

    public int UpdateCounter {
      get { return UpdateCounterParameter.Value.Value; }
      set { UpdateCounterParameter.Value.Value = value; }
    }

    public int UpdateInterval {
      get { return UpdateIntervalParameter.Value.Value; }
      set { UpdateIntervalParameter.Value.Value = value; }
    }
    #endregion

    public SymbolicDataAnalysisPoly10Analyzer() {
      #region Add parameters
      Parameters.Add(new LookupParameter<IDataAnalysisProblemData>(ProblemDataParameterName));
      Parameters.Add(new LookupParameter<ISymbolicDataAnalysisExpressionTreeInterpreter>(SymbolicDataAnalysisTreeInterpreterParameterName));
      Parameters.Add(new LookupParameter<IntValue>(GenerationsParameterName));
      Parameters.Add(new ValueParameter<DoubleValue>(PhenotypicSimilarityThresholdParameterName, "The phenotypic similarity threshold", new DoubleValue(0.9)));
      Parameters.Add(new ValueParameter<IntValue>(UpdateCounterParameterName, new IntValue(0)));
      Parameters.Add(new ValueParameter<IntValue>(UpdateIntervalParameterName, new IntValue(1)));
      #endregion
    }

    [StorableConstructor]
    protected SymbolicDataAnalysisPoly10Analyzer(bool deserializing)
      : base(deserializing) {
    }

    protected SymbolicDataAnalysisPoly10Analyzer(SymbolicDataAnalysisPoly10Analyzer original, Cloner cloner)
      : base(original, cloner) {
    }

    public override IDeepCloneable Clone(Cloner cloner) {
      return new SymbolicDataAnalysisPoly10Analyzer(this, cloner);
    }

    new public bool EnabledByDefault {
      get { return false; }
    }

    public override IOperation Apply() {
      #region Update counter & update interval
      UpdateCounter++;
      if (UpdateCounter != UpdateInterval) {
        return base.Apply();
      }
      UpdateCounter = 0;
      #endregion

      int generations = GenerationsParameter.ActualValue.Value;
      if (generations == 0)
        InitializeBuildingBlockCollection();

      var results = ResultCollectionParameter.ActualValue;
      var trees = SymbolicExpressionTreeParameter.ActualValue;
      var interpreter = (SymbolicDataAnalysisExpressionTreeLinearInterpreter)SymbolicDataAnalysisTreeInterpreterParameter.ActualValue;
      var dataset = ProblemDataParameter.ActualValue.Dataset;
      var rows = ProblemDataParameter.ActualValue.TrainingIndices.ToList();
      var bbFrequencies = evaluationMap.Keys.ToDictionary(x => x, x => 0);

      foreach (var key in evaluationMap.Keys) {
        var bb = fragmentMap[key];
        int len = bb.GetLength();
        foreach (var t in trees) {
          var root = t.Root.GetSubtree(0).GetSubtree(0);
          var nodes = root.IterateNodesPrefix().Where(x => x.GetLength() > len).ToList();

          for (int i = 0; i < nodes.Count; ++i) {
            var s = nodes[i];
            var values = interpreter.GetValues(s, dataset, rows);
            OnlineCalculatorError error;
            var r = OnlinePearsonsRCalculator.Calculate(values, evaluationMap[key], out error);
            var r2 = error == OnlineCalculatorError.None ? r * r : double.NaN;
            if (!double.IsNaN(r2) && r2 >= PhenotypicSimilarityThreshold) {
              bbFrequencies[key]++;
              i += s.GetLength();
            }
          }
        }
      }
      var table = (DataTable)results[BuildingBlocksFrequenciesTableName].Value;
      foreach (var pair in bbFrequencies) {
        var formatter = new SymbolicExpressionTreeStringFormatter();
        //        var label = formatter.Format(fragmentMap[pair.Key]) + "(" + prettyLabels[pair.Key] + ")";
        var label = prettyLabels[pair.Key];
        if (table.Rows.ContainsKey(label)) {
          var row = table.Rows[label];
          row.Values.Add(pair.Value);
        }
      }

      return base.Apply();
    }

    private void InitializeBuildingBlockCollection() {
      #region Add building blocks
      // building blocks
      const string x1 = "(variable 1 X1)";
      const string x2 = "(variable 1 X2)";
      const string x3 = "(variable 1 X3)";
      const string x4 = "(variable 1 X4)";
      const string x5 = "(variable 1 X5)";
      const string x6 = "(variable 1 X6)";
      const string x7 = "(variable 1 X7)";
      // x8 is never used in the formula
      // const string x8 = "(variable 1 X8)"; 
      const string x9 = "(variable 1 X9)";
      const string x10 = "(variable 1 X10)";
      string s1 = String.Format("(* {0} {1})", x1, x2);
      string s2 = String.Format("(* {0} {1})", x3, x4);
      string s3 = String.Format("(* {0} {1})", x5, x6);
      string s4 = String.Format("(* (* {0} {1}) {2})", x1, x7, x9);
      string s5 = String.Format("(* (* {0} {1}) {2})", x3, x6, x10);
      string s6 = String.Format("(+ {0} {1})", s1, s2); // x1x2 + x3x4
      string s7 = String.Format("(+ {0} {1})", s1, s3); // x1x2 + x5x6
      string s8 = String.Format("(+ {0} {1})", s2, s3); // x3x4 + x5x6
      string s9 = String.Format("(+ (+ {0} {1}) {2})", s1, s2, s3); // x1x2 + x3x4 + x5x6
      string s10 = String.Format("(+ (+ {0} {1}) {2})", s4, s5, s9); // x1x2 + x3x4 + x5x6 + x1x7x9 + x3x6x10
      prettyLabels[s1] = "X1*X2";
      prettyLabels[s2] = "X3*X4";
      prettyLabels[s3] = "X5*X6";
      prettyLabels[s4] = "X1*X7*X9";
      prettyLabels[s5] = "X3*X6*X10";
      prettyLabels[s6] = prettyLabels[s1] + " + " + prettyLabels[s2];
      prettyLabels[s7] = prettyLabels[s1] + " + " + prettyLabels[s3];
      prettyLabels[s8] = prettyLabels[s2] + " + " + prettyLabels[s3];
      prettyLabels[s9] = prettyLabels[s1] + " + " + prettyLabels[s2] + " + " + prettyLabels[s3];
      prettyLabels[s10] = prettyLabels[s9] + " + " + prettyLabels[s4] + " + " + prettyLabels[s5];
      #endregion
      var interpreter = SymbolicDataAnalysisTreeInterpreterParameter.ActualValue;
      var dataset = ProblemDataParameter.ActualValue.Dataset;
      var rows = ProblemDataParameter.ActualValue.TrainingIndices.ToList();

      foreach (var s in new[] { s1, s2, s3, s4, s5, s6, s7, s8, s9, s10 }) {
        if (evaluationMap.ContainsKey(s)) continue;
        var t = importer.Import(s);
        evaluationMap.Add(s, interpreter.GetSymbolicExpressionTreeValues(t, dataset, rows).ToList());
        fragmentMap.Add(s, t.Root.GetSubtree(0).GetSubtree(0));
      }

      var results = ResultCollectionParameter.ActualValue;
      DataTable table;
      if (!results.ContainsKey(BuildingBlocksFrequenciesTableName)) {
        table = new DataTable(BuildingBlocksFrequenciesTableName);
        results.Add(new Result(BuildingBlocksFrequenciesTableName, table));
      } else {
        table = (DataTable)results[BuildingBlocksFrequenciesTableName].Value;
      }
      table.Rows.Clear();
      foreach (var key in evaluationMap.Keys) {
        table.Rows.Add(new DataRow(prettyLabels[key]) { VisualProperties = { StartIndexZero = true } });
      }
    }
  }
}