source: branches/FitnessLandscapeAnalysis/HeuristicLab.Analysis.FitnessLandscape/BoxChart/RunCollectionMultiDiscretizer.cs @ 7403

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using HeuristicLab.Common;
using HeuristicLab.Core;
using HeuristicLab.Optimization;
using HeuristicLab.Parameters;
using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
using HeuristicLab.Data;

namespace HeuristicLab.Optimization.RunModification {

  [Item("RunCollection MultiDiscretizer",
    "Creates several levels from the distribution of several results accross a run collection and " +
    "assigns a discretized value. Non-existing numbers as well as NaN and infinities are excluded from the caluclation.")]
  [StorableClass]
  public class RunCollectionMultiDiscretizer : ParameterizedNamedItem, IRunCollectionModifier {

    public override bool CanChangeName { get { return false; } }
    public override bool CanChangeDescription { get { return false; } }

    #region Parameters
    public ValueParameter<CheckedItemList<StringValue>> SourceParameter {
      get { return (ValueParameter<CheckedItemList<StringValue>>)Parameters["Sources"]; }
    }
    public ValueParameter<StringValue> SuffixParameter {
      get { return (ValueParameter<StringValue>)Parameters["Suffix"]; }
    }
    public ValueParameter<DoubleValue> SpreadParameter {
      get { return (ValueParameter<DoubleValue>)Parameters["Spread"]; }
    }
    public ValueParameter<StringValue> GroupByParameter {
      get { return (ValueParameter<StringValue>)Parameters["GroupBy"]; }
    }    
    public ValueParameter<ItemList<StringValue>> LevelsParameter {
      get { return (ValueParameter<ItemList<StringValue>>)Parameters["Levels"]; }
    }    
    #endregion

    private IEnumerable<string> Sources { get { return SourceParameter.Value.CheckedItems.Select(i => i.Value.Value); } }
    private string Suffix { get { return SuffixParameter.Value.Value; } }
    private double Spread { get { return SpreadParameter.Value.Value; } }
    private string GroupBy { get { return GroupByParameter.Value.Value; } }
    private List<string> Levels { get { return LevelsParameter.Value.Select(v => v.Value).ToList(); } }

      #region Construction & Cloning
    [StorableConstructor]
    protected RunCollectionMultiDiscretizer(bool deserializing) : base(deserializing) { }
    protected RunCollectionMultiDiscretizer(RunCollectionMultiDiscretizer original, Cloner cloner) : base(original, cloner) { }
    public RunCollectionMultiDiscretizer() {
      Parameters.Add(new ValueParameter<CheckedItemList<StringValue>>("Sources", "Source value names to be discretized."));
      Parameters.Add(new ValueParameter<StringValue>("Suffix", "Suffix attached to new discretized values.", new StringValue("/Level")));
      Parameters.Add(new ValueParameter<DoubleValue>("Spread", "The number of standard deviations considered one additional level. Set to zero to use empirical distribution instead.", new DoubleValue(1)));
      Parameters.Add(new ValueParameter<StringValue>("GroupBy", "Create separate analyzes for different values of this variable.", new StringValue("")));
      Parameters.Add(new ValueParameter<ItemList<StringValue>>("Levels", "The list of levels to be assigned.",
        new ItemList<StringValue> {
          new StringValue("Very Low"),
          new StringValue("Low"),
          new StringValue("Average"),
          new StringValue("High"),
          new StringValue("Very High"),
        }));
    }
    public override IDeepCloneable Clone(Cloner cloner) {
      return new RunCollectionMultiDiscretizer(this, cloner);
    }
    #endregion

    #region IRunCollectionModifier Members

    public void Modify(List<IRun> runs) {
      foreach (var source in Sources) {
        var newName = source + Suffix;
        foreach (var group in runs
          .Select(r => new {Run = r, Value = GetValue(source, r)})
          .Where(r => r.Value.HasValue && !double.IsNaN(r.Value.Value) && !double.IsInfinity(r.Value.Value))
          .Select(r => new {r.Run, r.Value.Value, Bin = GetGroupByValue(r.Run)})
          .GroupBy(r => r.Bin).ToList()) {
          var values = group.Select(r => r.Value).ToList();
          if (values.Count > 0) {
            if (Spread > 0) {
              var avg = values.Average();
              var stdDev = values.StandardDeviation();
              foreach (var r in group) {
                r.Run.Results[newName] = new StringValue(Discretize(r.Value, avg, stdDev));
              }
            }
            else {
              values.Sort();
              var a = values.ToArray();
              foreach (var r in group) {
                r.Run.Results[newName] = new StringValue(Discretize(r.Value, a));
              }
            }
          }
        }
      }
    }

    private double? GetValue(string name, IRun run) {
      return CastValue(name, run.Results) ?? CastValue(name, run.Parameters);
    }

    private string GetGroupByValue(IRun run) {
      if (string.IsNullOrWhiteSpace(GroupBy))
        return String.Empty;
      IItem value;
      run.Results.TryGetValue(GroupBy, out value);
      if (value == null)
        run.Parameters.TryGetValue(GroupBy, out value);
      if (value != null)
        return value.ToString();
      else
        return String.Empty;
    }

    private double? CastValue(string name, IDictionary<string, IItem> variables) {
      IItem value;
      variables.TryGetValue(name, out value);
      var intValue = value as IntValue;
      if (intValue != null) 
        return intValue.Value;
      var doubleValue = value as DoubleValue;
      if (doubleValue != null)
        return doubleValue.Value;
      return null;
    }

    private string Discretize(double value, double avg, double stdDev) {
      double dev = (value - avg)/(stdDev*Spread);
      int index;
      if (Levels.Count % 2 == 1) {
        index = (int) Math.Floor(Math.Abs(dev));
        index = (Levels.Count - 1)/2 + Math.Sign(dev) * index;
      } else {
        index = (int) Math.Ceiling(Math.Abs(dev));
        if (dev > 0)
          index = Levels.Count/2 + index;
        else
          index = Levels.Count/2 + 1 - index;
      }
      return Levels[Math.Min(Levels.Count - 1, Math.Max(0, index))];
    }

    private string Discretize(double value, double[] values) {
      var index = Array.BinarySearch(values, value);
      var pos = 1.0*(index < 0 ? ~index : index)/(values.Length-1);
      return Levels[Math.Min(Levels.Count - 1, Math.Max(0, (int) Math.Round(pos*(Levels.Count-1))))];
    }

    #endregion
  }
}