Free cookie consent management tool by TermsFeed Policy Generator

source: trunk/sources/HeuristicLab.Problems.DataAnalysis/3.4/Implementation/Classification/ClassificationProblemData.cs @ 15612

Last change on this file since 15612 was 15583, checked in by swagner, 7 years ago

#2640: Updated year of copyrights in license headers
File size: 25.6 KB
RevLine 
[5559]1#region License Information
2/* HeuristicLab
[15583]3 * Copyright (C) 2002-2018 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
[5559]4 *
5 * This file is part of HeuristicLab.
6 *
7 * HeuristicLab is free software: you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation, either version 3 of the License, or
10 * (at your option) any later version.
11 *
12 * HeuristicLab is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
15 * GNU General Public License for more details.
16 *
17 * You should have received a copy of the GNU General Public License
18 * along with HeuristicLab. If not, see <http://www.gnu.org/licenses/>.
19 */
20#endregion
21
22using System;
23using System.Collections.Generic;
24using System.Linq;
25using HeuristicLab.Common;
[5601]26using HeuristicLab.Core;
27using HeuristicLab.Data;
28using HeuristicLab.Parameters;
[5559]29using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
30
31namespace HeuristicLab.Problems.DataAnalysis {
32  [StorableClass]
[5601]33  [Item("ClassificationProblemData", "Represents an item containing all data defining a classification problem.")]
[7134]34  public class ClassificationProblemData : DataAnalysisProblemData, IClassificationProblemData, IStorableContent {
[6666]35    protected const string TargetVariableParameterName = "TargetVariable";
36    protected const string ClassNamesParameterName = "ClassNames";
37    protected const string ClassificationPenaltiesParameterName = "ClassificationPenalties";
[11766]38    protected const string PositiveClassParameterName = "PositiveClass";
[7266]39    protected const int MaximumNumberOfClasses = 100;
40    protected const int InspectedRowsToDetermineTargets = 2000;
[5601]41
[7134]42    public string Filename { get; set; }
43
[5559]44    #region default data
45    private static string[] defaultVariableNames = new string[] { "sample", "clump thickness", "cell size", "cell shape", "marginal adhesion", "epithelial cell size", "bare nuclei", "chromatin", "nucleoli", "mitoses", "class" };
46    private static double[,] defaultData = new double[,]{
47     {1000025,5,1,1,1,2,1,3,1,1,2      },
48     {1002945,5,4,4,5,7,10,3,2,1,2     },
49     {1015425,3,1,1,1,2,2,3,1,1,2      },
50     {1016277,6,8,8,1,3,4,3,7,1,2      },
51     {1017023,4,1,1,3,2,1,3,1,1,2      },
52     {1017122,8,10,10,8,7,10,9,7,1,4   },
53     {1018099,1,1,1,1,2,10,3,1,1,2     },
54     {1018561,2,1,2,1,2,1,3,1,1,2      },
55     {1033078,2,1,1,1,2,1,1,1,5,2      },
56     {1033078,4,2,1,1,2,1,2,1,1,2      },
57     {1035283,1,1,1,1,1,1,3,1,1,2      },
58     {1036172,2,1,1,1,2,1,2,1,1,2      },
59     {1041801,5,3,3,3,2,3,4,4,1,4      },
60     {1043999,1,1,1,1,2,3,3,1,1,2      },
61     {1044572,8,7,5,10,7,9,5,5,4,4     },
62     {1047630,7,4,6,4,6,1,4,3,1,4      },
63     {1048672,4,1,1,1,2,1,2,1,1,2      },
64     {1049815,4,1,1,1,2,1,3,1,1,2      },
65     {1050670,10,7,7,6,4,10,4,1,2,4    },
66     {1050718,6,1,1,1,2,1,3,1,1,2      },
67     {1054590,7,3,2,10,5,10,5,4,4,4    },
68     {1054593,10,5,5,3,6,7,7,10,1,4    },
69     {1056784,3,1,1,1,2,1,2,1,1,2      },
70     {1057013,8,4,5,1,2,2,7,3,1,4      },
71     {1059552,1,1,1,1,2,1,3,1,1,2      },
72     {1065726,5,2,3,4,2,7,3,6,1,4      },
73     {1066373,3,2,1,1,1,1,2,1,1,2      },
74     {1066979,5,1,1,1,2,1,2,1,1,2      },
75     {1067444,2,1,1,1,2,1,2,1,1,2      },
76     {1070935,1,1,3,1,2,1,1,1,1,2      },
77     {1070935,3,1,1,1,1,1,2,1,1,2      },
78     {1071760,2,1,1,1,2,1,3,1,1,2      },
79     {1072179,10,7,7,3,8,5,7,4,3,4     },
80     {1074610,2,1,1,2,2,1,3,1,1,2      },
81     {1075123,3,1,2,1,2,1,2,1,1,2      },
82     {1079304,2,1,1,1,2,1,2,1,1,2      },
83     {1080185,10,10,10,8,6,1,8,9,1,4   },
84     {1081791,6,2,1,1,1,1,7,1,1,2      },
85     {1084584,5,4,4,9,2,10,5,6,1,4     },
86     {1091262,2,5,3,3,6,7,7,5,1,4      },
87     {1096800,6,6,6,9,6,4,7,8,1,2      },
88     {1099510,10,4,3,1,3,3,6,5,2,4     },
89     {1100524,6,10,10,2,8,10,7,3,3,4   },
90     {1102573,5,6,5,6,10,1,3,1,1,4     },
91     {1103608,10,10,10,4,8,1,8,10,1,4  },
92     {1103722,1,1,1,1,2,1,2,1,2,2      },
93     {1105257,3,7,7,4,4,9,4,8,1,4      },
94     {1105524,1,1,1,1,2,1,2,1,1,2      },
95     {1106095,4,1,1,3,2,1,3,1,1,2      },
96     {1106829,7,8,7,2,4,8,3,8,2,4      },
97     {1108370,9,5,8,1,2,3,2,1,5,4      },
98     {1108449,5,3,3,4,2,4,3,4,1,4      },
99     {1110102,10,3,6,2,3,5,4,10,2,4    },
100     {1110503,5,5,5,8,10,8,7,3,7,4     },
101     {1110524,10,5,5,6,8,8,7,1,1,4     },
102     {1111249,10,6,6,3,4,5,3,6,1,4     },
103     {1112209,8,10,10,1,3,6,3,9,1,4    },
104     {1113038,8,2,4,1,5,1,5,4,4,4      },
105     {1113483,5,2,3,1,6,10,5,1,1,4     },
106     {1113906,9,5,5,2,2,2,5,1,1,4      },
107     {1115282,5,3,5,5,3,3,4,10,1,4     },
108     {1115293,1,1,1,1,2,2,2,1,1,2      },
109     {1116116,9,10,10,1,10,8,3,3,1,4   },
110     {1116132,6,3,4,1,5,2,3,9,1,4      },
111     {1116192,1,1,1,1,2,1,2,1,1,2      },
112     {1116998,10,4,2,1,3,2,4,3,10,4    },
113     {1117152,4,1,1,1,2,1,3,1,1,2      },
114     {1118039,5,3,4,1,8,10,4,9,1,4     },
115     {1120559,8,3,8,3,4,9,8,9,8,4      },
116     {1121732,1,1,1,1,2,1,3,2,1,2      },
117     {1121919,5,1,3,1,2,1,2,1,1,2      },
118     {1123061,6,10,2,8,10,2,7,8,10,4   },
119     {1124651,1,3,3,2,2,1,7,2,1,2      },
120     {1125035,9,4,5,10,6,10,4,8,1,4    },
121     {1126417,10,6,4,1,3,4,3,2,3,4     },
122     {1131294,1,1,2,1,2,2,4,2,1,2      },
123     {1132347,1,1,4,1,2,1,2,1,1,2      },
124     {1133041,5,3,1,2,2,1,2,1,1,2      },
125     {1133136,3,1,1,1,2,3,3,1,1,2      },
126     {1136142,2,1,1,1,3,1,2,1,1,2      },
127     {1137156,2,2,2,1,1,1,7,1,1,2      },
128     {1143978,4,1,1,2,2,1,2,1,1,2      },
129     {1143978,5,2,1,1,2,1,3,1,1,2      },
130     {1147044,3,1,1,1,2,2,7,1,1,2      },
131     {1147699,3,5,7,8,8,9,7,10,7,4     },
132     {1147748,5,10,6,1,10,4,4,10,10,4  },
133     {1148278,3,3,6,4,5,8,4,4,1,4      },
134     {1148873,3,6,6,6,5,10,6,8,3,4     },
135     {1152331,4,1,1,1,2,1,3,1,1,2      },
136     {1155546,2,1,1,2,3,1,2,1,1,2      },
137     {1156272,1,1,1,1,2,1,3,1,1,2      },
138     {1156948,3,1,1,2,2,1,1,1,1,2      },
139     {1157734,4,1,1,1,2,1,3,1,1,2      },
140     {1158247,1,1,1,1,2,1,2,1,1,2      },
141     {1160476,2,1,1,1,2,1,3,1,1,2      },
142     {1164066,1,1,1,1,2,1,3,1,1,2      },
143     {1165297,2,1,1,2,2,1,1,1,1,2      },
144     {1165790,5,1,1,1,2,1,3,1,1,2      },
145     {1165926,9,6,9,2,10,6,2,9,10,4    },
146     {1166630,7,5,6,10,5,10,7,9,4,4    },
147     {1166654,10,3,5,1,10,5,3,10,2,4   },
148     {1167439,2,3,4,4,2,5,2,5,1,4      },
149     {1167471,4,1,2,1,2,1,3,1,1,2      },
150     {1168359,8,2,3,1,6,3,7,1,1,4      },
151     {1168736,10,10,10,10,10,1,8,8,8,4 },
152     {1169049,7,3,4,4,3,3,3,2,7,4      },
153     {1170419,10,10,10,8,2,10,4,1,1,4  },
154     {1170420,1,6,8,10,8,10,5,7,1,4    },
155     {1171710,1,1,1,1,2,1,2,3,1,2      },
156     {1171710,6,5,4,4,3,9,7,8,3,4      },
157     {1171795,1,3,1,2,2,2,5,3,2,2      },
158     {1171845,8,6,4,3,5,9,3,1,1,4      },
159     {1172152,10,3,3,10,2,10,7,3,3,4   },
160     {1173216,10,10,10,3,10,8,8,1,1,4  },
161     {1173235,3,3,2,1,2,3,3,1,1,2      },
162     {1173347,1,1,1,1,2,5,1,1,1,2      },
163     {1173347,8,3,3,1,2,2,3,2,1,2      },
164     {1173509,4,5,5,10,4,10,7,5,8,4    },
165     {1173514,1,1,1,1,4,3,1,1,1,2      },
166     {1173681,3,2,1,1,2,2,3,1,1,2      },
167     {1174057,1,1,2,2,2,1,3,1,1,2      },
168     {1174057,4,2,1,1,2,2,3,1,1,2      },
169     {1174131,10,10,10,2,10,10,5,3,3,4 },
170     {1174428,5,3,5,1,8,10,5,3,1,4     },
171     {1175937,5,4,6,7,9,7,8,10,1,4     },
172     {1176406,1,1,1,1,2,1,2,1,1,2      },
173     {1176881,7,5,3,7,4,10,7,5,5,4        }
174};
[6672]175    private static readonly Dataset defaultDataset;
176    private static readonly IEnumerable<string> defaultAllowedInputVariables;
177    private static readonly string defaultTargetVariable;
[6666]178
[6672]179    private static readonly ClassificationProblemData emptyProblemData;
[6666]180    public static ClassificationProblemData EmptyProblemData {
181      get { return EmptyProblemData; }
182    }
183
[5559]184    static ClassificationProblemData() {
185      defaultDataset = new Dataset(defaultVariableNames, defaultData);
186      defaultDataset.Name = "Wisconsin classification problem";
187      defaultDataset.Description = "subset from to ..";
188
189      defaultAllowedInputVariables = defaultVariableNames.Except(new List<string>() { "sample", "class" });
190      defaultTargetVariable = "class";
[6666]191
192      var problemData = new ClassificationProblemData();
193      problemData.Parameters.Clear();
194      problemData.Name = "Empty Classification ProblemData";
195      problemData.Description = "This ProblemData acts as place holder before the correct problem data is loaded.";
196      problemData.isEmpty = true;
197
198      problemData.Parameters.Add(new FixedValueParameter<Dataset>(DatasetParameterName, "", new Dataset()));
199      problemData.Parameters.Add(new FixedValueParameter<ReadOnlyCheckedItemList<StringValue>>(InputVariablesParameterName, ""));
200      problemData.Parameters.Add(new FixedValueParameter<IntRange>(TrainingPartitionParameterName, "", (IntRange)new IntRange(0, 0).AsReadOnly()));
201      problemData.Parameters.Add(new FixedValueParameter<IntRange>(TestPartitionParameterName, "", (IntRange)new IntRange(0, 0).AsReadOnly()));
202      problemData.Parameters.Add(new ConstrainedValueParameter<StringValue>(TargetVariableParameterName, new ItemSet<StringValue>()));
203      problemData.Parameters.Add(new FixedValueParameter<StringMatrix>(ClassNamesParameterName, "", new StringMatrix(0, 0).AsReadOnly()));
204      problemData.Parameters.Add(new FixedValueParameter<DoubleMatrix>(ClassificationPenaltiesParameterName, "", (DoubleMatrix)new DoubleMatrix(0, 0).AsReadOnly()));
205      emptyProblemData = problemData;
[5559]206    }
207    #endregion
208
[5601]209    #region parameter properties
[8121]210    public IConstrainedValueParameter<StringValue> TargetVariableParameter {
211      get { return (IConstrainedValueParameter<StringValue>)Parameters[TargetVariableParameterName]; }
[5601]212    }
213    public IFixedValueParameter<StringMatrix> ClassNamesParameter {
214      get { return (IFixedValueParameter<StringMatrix>)Parameters[ClassNamesParameterName]; }
215    }
[11766]216    public IConstrainedValueParameter<StringValue> PositiveClassParameter {
217      get { return (IConstrainedValueParameter<StringValue>)Parameters[PositiveClassParameterName]; }
[11763]218    }
[5601]219    public IFixedValueParameter<DoubleMatrix> ClassificationPenaltiesParameter {
220      get { return (IFixedValueParameter<DoubleMatrix>)Parameters[ClassificationPenaltiesParameterName]; }
221    }
222    #endregion
223
[5649]224    #region properties
[5559]225    public string TargetVariable {
[5601]226      get { return TargetVariableParameter.Value.Value; }
[10540]227      set {
228        if (value == null) throw new ArgumentNullException("targetVariable", "The provided value for the targetVariable is null.");
229        if (value == TargetVariable) return;
230
231
232        var matchingParameterValue = TargetVariableParameter.ValidValues.FirstOrDefault(v => v.Value == value);
233        if (matchingParameterValue == null) throw new ArgumentException("The provided value is not valid as the targetVariable.", "targetVariable");
234        TargetVariableParameter.Value = matchingParameterValue;
235      }
[5601]236    }
[5559]237
[8554]238    private List<double> classValuesCache;
239    private List<double> ClassValuesCache {
[5601]240      get {
[8554]241        if (classValuesCache == null) {
242          classValuesCache = Dataset.GetDoubleValues(TargetVariableParameter.Value.Value).Distinct().OrderBy(x => x).ToList();
[5559]243        }
[8554]244        return classValuesCache;
[5559]245      }
246    }
[8554]247    public IEnumerable<double> ClassValues {
248      get { return ClassValuesCache; }
[5559]249    }
250    public int Classes {
[8554]251      get { return ClassValuesCache.Count; }
[5559]252    }
253
[8554]254    private List<string> classNamesCache;
255    private List<string> ClassNamesCache {
[5601]256      get {
[8554]257        if (classNamesCache == null) {
258          classNamesCache = new List<string>();
[5601]259          for (int i = 0; i < ClassNamesParameter.Value.Rows; i++)
[8554]260            classNamesCache.Add(ClassNamesParameter.Value[i, 0]);
[5601]261        }
[8554]262        return classNamesCache;
[5559]263      }
264    }
[8554]265    public IEnumerable<string> ClassNames {
266      get { return ClassNamesCache; }
[5559]267    }
[11763]268
[11766]269    public string PositiveClass {
270      get { return PositiveClassParameter.Value.Value; }
[11763]271      set {
[11766]272        var matchingValue = PositiveClassParameter.ValidValues.SingleOrDefault(x => x.Value == value);
[11763]273        if (matchingValue == null) throw new ArgumentException(string.Format("{0} cannot be set as positive class.", value));
[11766]274        PositiveClassParameter.Value = matchingValue;
[11763]275      }
276    }
[5559]277    #endregion
278
279
280    [StorableConstructor]
281    protected ClassificationProblemData(bool deserializing) : base(deserializing) { }
[5601]282    [StorableHook(HookType.AfterDeserialization)]
283    private void AfterDeserialization() {
284      RegisterParameterEvents();
[13446]285
286      classNamesCache = new List<string>();
287      for (int i = 0; i < ClassNamesParameter.Value.Rows; i++)
288        classNamesCache.Add(ClassNamesParameter.Value[i, 0]);
289
[11763]290      // BackwardsCompatibility3.4
291      #region Backwards compatible code, remove with 3.5
[11766]292      if (!Parameters.ContainsKey(PositiveClassParameterName)) {
[11763]293        var validValues = new ItemSet<StringValue>(ClassNames.Select(s => new StringValue(s).AsReadOnly()));
[11766]294        Parameters.Add(new ConstrainedValueParameter<StringValue>(PositiveClassParameterName,
[11763]295          "The positive class which is used for quality measure calculation (e.g., specifity, sensitivity,...)", validValues, validValues.First()));
296      }
297      #endregion
298
[5601]299    }
[5559]300
[5601]301    protected ClassificationProblemData(ClassificationProblemData original, Cloner cloner)
302      : base(original, cloner) {
303      RegisterParameterEvents();
[13446]304      classNamesCache = new List<string>();
305      for (int i = 0; i < ClassNamesParameter.Value.Rows; i++)
306        classNamesCache.Add(ClassNamesParameter.Value[i, 0]);
[5559]307    }
[6666]308    public override IDeepCloneable Clone(Cloner cloner) {
309      if (this == emptyProblemData) return emptyProblemData;
310      return new ClassificationProblemData(this, cloner);
311    }
[5559]312
[5601]313    public ClassificationProblemData() : this(defaultDataset, defaultAllowedInputVariables, defaultTargetVariable) { }
[8528]314
315    public ClassificationProblemData(IClassificationProblemData classificationProblemData)
316      : this(classificationProblemData.Dataset, classificationProblemData.AllowedInputVariables, classificationProblemData.TargetVariable) {
317      TrainingPartition.Start = classificationProblemData.TrainingPartition.Start;
318      TrainingPartition.End = classificationProblemData.TrainingPartition.End;
319      TestPartition.Start = classificationProblemData.TestPartition.Start;
320      TestPartition.End = classificationProblemData.TestPartition.End;
[8716]321
322      for (int i = 0; i < classificationProblemData.ClassNames.Count(); i++)
323        ClassNamesParameter.Value[i, 0] = classificationProblemData.ClassNames.ElementAt(i);
[8717]324
[15517]325      //mkommend: The positive class depends on the class names and as a result must only be set after the classe names parameter.
[15498]326      PositiveClass = classificationProblemData.PositiveClass;
327
[8717]328      for (int i = 0; i < Classes; i++) {
329        for (int j = 0; j < Classes; j++) {
[8745]330          ClassificationPenaltiesParameter.Value[i, j] = classificationProblemData.GetClassificationPenalty(ClassValuesCache[i], ClassValuesCache[j]);
[8717]331        }
332      }
[8528]333    }
334
[12509]335    public ClassificationProblemData(IDataset dataset, IEnumerable<string> allowedInputVariables, string targetVariable, IEnumerable<ITransformation> transformations = null)
[11114]336      : base(dataset, allowedInputVariables, transformations ?? Enumerable.Empty<ITransformation>()) {
[6186]337      var validTargetVariableValues = CheckVariablesForPossibleTargetVariables(dataset).Select(x => new StringValue(x).AsReadOnly()).ToList();
338      var target = validTargetVariableValues.Where(x => x.Value == targetVariable).DefaultIfEmpty(validTargetVariableValues.First()).First();
339
340      Parameters.Add(new ConstrainedValueParameter<StringValue>(TargetVariableParameterName, new ItemSet<StringValue>(validTargetVariableValues), target));
[5847]341      Parameters.Add(new FixedValueParameter<StringMatrix>(ClassNamesParameterName, ""));
[11766]342      Parameters.Add(new ConstrainedValueParameter<StringValue>(PositiveClassParameterName, "The positive class which is used for quality measure calculation (e.g., specifity, sensitivity,...)"));
[5847]343      Parameters.Add(new FixedValueParameter<DoubleMatrix>(ClassificationPenaltiesParameterName, ""));
[5559]344
[8802]345      RegisterParameterEvents();
[5601]346      ResetTargetVariableDependentMembers();
[5559]347    }
348
[12509]349    public static IEnumerable<string> CheckVariablesForPossibleTargetVariables(IDataset dataset) {
[6223]350      int maxSamples = Math.Min(InspectedRowsToDetermineTargets, dataset.Rows);
[6740]351      var validTargetVariables = (from v in dataset.DoubleVariables
352                                  let distinctValues = dataset.GetDoubleValues(v)
[6654]353                                    .Take(maxSamples)
354                                    .Distinct()
355                                    .Count()
[9449]356                                  where distinctValues <= MaximumNumberOfClasses
[6654]357                                  select v).ToArray();
[6186]358
359      if (!validTargetVariables.Any())
[6223]360        throw new ArgumentException("Import of classification problem data was not successful, because no target variable was found." +
361          " A target variable must have at most " + MaximumNumberOfClasses + " distinct values to be applicable to classification.");
[6186]362      return validTargetVariables;
363    }
364
365
[5601]366    private void ResetTargetVariableDependentMembers() {
[6654]367      DeregisterParameterEvents();
[5559]368
[5601]369      ((IStringConvertibleMatrix)ClassNamesParameter.Value).Columns = 1;
[8554]370      ((IStringConvertibleMatrix)ClassNamesParameter.Value).Rows = ClassValuesCache.Count;
[5601]371      for (int i = 0; i < Classes; i++)
[8554]372        ClassNamesParameter.Value[i, 0] = "Class " + ClassValuesCache[i];
[5601]373      ClassNamesParameter.Value.ColumnNames = new List<string>() { "ClassNames" };
374      ClassNamesParameter.Value.RowNames = ClassValues.Select(s => "ClassValue: " + s);
[5559]375
[11766]376      PositiveClassParameter.ValidValues.Clear();
[11763]377      foreach (var className in ClassNames) {
[11766]378        PositiveClassParameter.ValidValues.Add(new StringValue(className).AsReadOnly());
[11763]379      }
380
[5601]381      ((IStringConvertibleMatrix)ClassificationPenaltiesParameter.Value).Rows = Classes;
382      ((IStringConvertibleMatrix)ClassificationPenaltiesParameter.Value).Columns = Classes;
383      ClassificationPenaltiesParameter.Value.RowNames = ClassNames.Select(name => "Actual " + name);
384      ClassificationPenaltiesParameter.Value.ColumnNames = ClassNames.Select(name => "Estimated " + name);
385      for (int i = 0; i < Classes; i++) {
386        for (int j = 0; j < Classes; j++) {
387          if (i != j) ClassificationPenaltiesParameter.Value[i, j] = 1;
388          else ClassificationPenaltiesParameter.Value[i, j] = 0;
[5559]389        }
390      }
[5601]391      RegisterParameterEvents();
[5559]392    }
393
394    public string GetClassName(double classValue) {
[8554]395      if (!ClassValuesCache.Contains(classValue)) throw new ArgumentException();
396      int index = ClassValuesCache.IndexOf(classValue);
397      return ClassNamesCache[index];
[5559]398    }
399    public double GetClassValue(string className) {
[8554]400      if (!ClassNamesCache.Contains(className)) throw new ArgumentException();
401      int index = ClassNamesCache.IndexOf(className);
402      return ClassValuesCache[index];
[5559]403    }
404    public void SetClassName(double classValue, string className) {
[8554]405      if (!ClassValuesCache.Contains(classValue)) throw new ArgumentException();
406      int index = ClassValuesCache.IndexOf(classValue);
[5601]407      ClassNamesParameter.Value[index, 0] = className;
[8554]408      // updating of class names cache is not necessary here as the parameter value fires a changed event which updates the cache
[5559]409    }
410
411    public double GetClassificationPenalty(string correctClassName, string estimatedClassName) {
412      return GetClassificationPenalty(GetClassValue(correctClassName), GetClassValue(estimatedClassName));
413    }
414    public double GetClassificationPenalty(double correctClassValue, double estimatedClassValue) {
[8554]415      int correctClassIndex = ClassValuesCache.IndexOf(correctClassValue);
416      int estimatedClassIndex = ClassValuesCache.IndexOf(estimatedClassValue);
417      return ClassificationPenaltiesParameter.Value[correctClassIndex, estimatedClassIndex];
[5559]418    }
419    public void SetClassificationPenalty(string correctClassName, string estimatedClassName, double penalty) {
420      SetClassificationPenalty(GetClassValue(correctClassName), GetClassValue(estimatedClassName), penalty);
421    }
422    public void SetClassificationPenalty(double correctClassValue, double estimatedClassValue, double penalty) {
[8554]423      int correctClassIndex = ClassValuesCache.IndexOf(correctClassValue);
424      int estimatedClassIndex = ClassValuesCache.IndexOf(estimatedClassValue);
[5601]425
426      ClassificationPenaltiesParameter.Value[correctClassIndex, estimatedClassIndex] = penalty;
[5559]427    }
428
[5601]429    #region events
430    private void RegisterParameterEvents() {
431      TargetVariableParameter.ValueChanged += new EventHandler(TargetVariableParameter_ValueChanged);
432      ClassNamesParameter.Value.Reset += new EventHandler(Parameter_ValueChanged);
[8716]433      ClassNamesParameter.Value.ItemChanged += new EventHandler<EventArgs<int, int>>(Parameter_ValueChanged);
[8717]434      ClassificationPenaltiesParameter.Value.ItemChanged += new EventHandler<EventArgs<int, int>>(Parameter_ValueChanged);
435      ClassificationPenaltiesParameter.Value.Reset += new EventHandler(Parameter_ValueChanged);
[5559]436    }
[6654]437    private void DeregisterParameterEvents() {
[5601]438      TargetVariableParameter.ValueChanged -= new EventHandler(TargetVariableParameter_ValueChanged);
439      ClassNamesParameter.Value.Reset -= new EventHandler(Parameter_ValueChanged);
[8716]440      ClassNamesParameter.Value.ItemChanged -= new EventHandler<EventArgs<int, int>>(Parameter_ValueChanged);
[8717]441      ClassificationPenaltiesParameter.Value.ItemChanged -= new EventHandler<EventArgs<int, int>>(Parameter_ValueChanged);
442      ClassificationPenaltiesParameter.Value.Reset -= new EventHandler(Parameter_ValueChanged);
[5559]443    }
[5601]444
445    private void TargetVariableParameter_ValueChanged(object sender, EventArgs e) {
[8554]446      classValuesCache = null;
447      classNamesCache = null;
[5601]448      ResetTargetVariableDependentMembers();
449      OnChanged();
450    }
451    private void Parameter_ValueChanged(object sender, EventArgs e) {
[11766]452      var oldPositiveClass = PositiveClass;
[11763]453      var oldClassNames = classNamesCache;
454      var index = oldClassNames.IndexOf(oldPositiveClass);
455
[8554]456      classNamesCache = null;
[8717]457      ClassificationPenaltiesParameter.Value.RowNames = ClassNames.Select(name => "Actual " + name);
458      ClassificationPenaltiesParameter.Value.ColumnNames = ClassNames.Select(name => "Estimated " + name);
[11763]459
[11766]460      PositiveClassParameter.ValidValues.Clear();
[11763]461      foreach (var className in ClassNames) {
[11766]462        PositiveClassParameter.ValidValues.Add(new StringValue(className).AsReadOnly());
[11763]463      }
[11766]464      PositiveClassParameter.Value = PositiveClassParameter.ValidValues.ElementAt(index);
[11763]465
[5601]466      OnChanged();
467    }
468    #endregion
[10540]469
470    protected override bool IsProblemDataCompatible(IDataAnalysisProblemData problemData, out string errorMessage) {
471      if (problemData == null) throw new ArgumentNullException("problemData", "The provided problemData is null.");
472      IClassificationProblemData classificationProblemData = problemData as IClassificationProblemData;
473      if (classificationProblemData == null)
474        throw new ArgumentException("The problem data is no classification problem data. Instead a " + problemData.GetType().GetPrettyName() + " was provided.", "problemData");
475
476      var returnValue = base.IsProblemDataCompatible(classificationProblemData, out errorMessage);
477      //check targetVariable
478      if (classificationProblemData.InputVariables.All(var => var.Value != TargetVariable)) {
479        errorMessage = string.Format("The target variable {0} is not present in the new problem data.", TargetVariable)
480                       + Environment.NewLine + errorMessage;
481        return false;
482      }
483
484      var newClassValues = classificationProblemData.Dataset.GetDoubleValues(TargetVariable).Distinct().OrderBy(x => x);
485      if (!newClassValues.SequenceEqual(ClassValues)) {
486        errorMessage = errorMessage + string.Format("The class values differ in the provided classification problem data.");
[11763]487        returnValue = false;
[10540]488      }
489
[11766]490      var newPositivieClassName = classificationProblemData.PositiveClass;
491      if (newPositivieClassName != PositiveClass) {
[11763]492        errorMessage = errorMessage + string.Format("The positive class differs in the provided classification problem data.");
493        returnValue = false;
494      }
495
[10540]496      return returnValue;
497    }
498
499    public override void AdjustProblemDataProperties(IDataAnalysisProblemData problemData) {
500      if (problemData == null) throw new ArgumentNullException("problemData", "The provided problemData is null.");
501      ClassificationProblemData classificationProblemData = problemData as ClassificationProblemData;
502      if (classificationProblemData == null)
503        throw new ArgumentException("The problem data is not a classification problem data. Instead a " + problemData.GetType().GetPrettyName() + " was provided.", "problemData");
504
505      base.AdjustProblemDataProperties(problemData);
506      TargetVariable = classificationProblemData.TargetVariable;
507      for (int i = 0; i < classificationProblemData.ClassNames.Count(); i++)
508        ClassNamesParameter.Value[i, 0] = classificationProblemData.ClassNames.ElementAt(i);
509
[11766]510      PositiveClass = classificationProblemData.PositiveClass;
[11763]511
[10540]512      for (int i = 0; i < Classes; i++) {
513        for (int j = 0; j < Classes; j++) {
514          ClassificationPenaltiesParameter.Value[i, j] = classificationProblemData.GetClassificationPenalty(ClassValuesCache[i], ClassValuesCache[j]);
515        }
516      }
517    }
[5559]518  }
519}
Note: See TracBrowser for help on using the repository browser.