source: trunk/sources/HeuristicLab.Problems.DataAnalysis/3.4/Implementation/Classification/ClassificationProblemData.cs @ 7134

Last change on this file since 7134 was 7134, checked in by mkommend, 10 years ago

#1698: Implemented IStorableContent in all ProblemDatas.

File size: 19.7 KB
Line 
1#region License Information
2/* HeuristicLab
3 * Copyright (C) 2002-2011 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
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.IO;
25using System.Linq;
26using HeuristicLab.Common;
27using HeuristicLab.Core;
28using HeuristicLab.Data;
29using HeuristicLab.Parameters;
30using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
31
32namespace HeuristicLab.Problems.DataAnalysis {
33  [StorableClass]
34  [Item("ClassificationProblemData", "Represents an item containing all data defining a classification problem.")]
35  public class ClassificationProblemData : DataAnalysisProblemData, IClassificationProblemData, IStorableContent {
36    protected const string TargetVariableParameterName = "TargetVariable";
37    protected const string ClassNamesParameterName = "ClassNames";
38    protected const string ClassificationPenaltiesParameterName = "ClassificationPenalties";
39    protected const int MaximumNumberOfClasses = 20;
40    protected const int InspectedRowsToDetermineTargets = 500;
41
42    public string Filename { get; set; }
43
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};
175    private static readonly Dataset defaultDataset;
176    private static readonly IEnumerable<string> defaultAllowedInputVariables;
177    private static readonly string defaultTargetVariable;
178
179    private static readonly ClassificationProblemData emptyProblemData;
180    public static ClassificationProblemData EmptyProblemData {
181      get { return EmptyProblemData; }
182    }
183
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";
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;
206    }
207    #endregion
208
209    #region parameter properties
210    public ConstrainedValueParameter<StringValue> TargetVariableParameter {
211      get { return (ConstrainedValueParameter<StringValue>)Parameters[TargetVariableParameterName]; }
212    }
213    public IFixedValueParameter<StringMatrix> ClassNamesParameter {
214      get { return (IFixedValueParameter<StringMatrix>)Parameters[ClassNamesParameterName]; }
215    }
216    public IFixedValueParameter<DoubleMatrix> ClassificationPenaltiesParameter {
217      get { return (IFixedValueParameter<DoubleMatrix>)Parameters[ClassificationPenaltiesParameterName]; }
218    }
219    #endregion
220
221    #region properties
222    public string TargetVariable {
223      get { return TargetVariableParameter.Value.Value; }
224    }
225
226    private List<double> classValues;
227    public List<double> ClassValues {
228      get {
229        if (classValues == null) {
230          classValues = Dataset.GetDoubleValues(TargetVariableParameter.Value.Value).Distinct().ToList();
231          classValues.Sort();
232        }
233        return classValues;
234      }
235    }
236    IEnumerable<double> IClassificationProblemData.ClassValues {
237      get { return ClassValues; }
238    }
239
240    public int Classes {
241      get { return ClassValues.Count; }
242    }
243
244    private List<string> classNames;
245    public List<string> ClassNames {
246      get {
247        if (classNames == null) {
248          classNames = new List<string>();
249          for (int i = 0; i < ClassNamesParameter.Value.Rows; i++)
250            classNames.Add(ClassNamesParameter.Value[i, 0]);
251        }
252        return classNames;
253      }
254    }
255    IEnumerable<string> IClassificationProblemData.ClassNames {
256      get { return ClassNames; }
257    }
258
259    private Dictionary<Tuple<double, double>, double> classificationPenaltiesCache = new Dictionary<Tuple<double, double>, double>();
260    #endregion
261
262
263    [StorableConstructor]
264    protected ClassificationProblemData(bool deserializing) : base(deserializing) { }
265    [StorableHook(HookType.AfterDeserialization)]
266    private void AfterDeserialization() {
267      RegisterParameterEvents();
268    }
269
270    protected ClassificationProblemData(ClassificationProblemData original, Cloner cloner)
271      : base(original, cloner) {
272      RegisterParameterEvents();
273    }
274    public override IDeepCloneable Clone(Cloner cloner) {
275      if (this == emptyProblemData) return emptyProblemData;
276      return new ClassificationProblemData(this, cloner);
277    }
278
279    public ClassificationProblemData() : this(defaultDataset, defaultAllowedInputVariables, defaultTargetVariable) { }
280    public ClassificationProblemData(Dataset dataset, IEnumerable<string> allowedInputVariables, string targetVariable)
281      : base(dataset, allowedInputVariables) {
282      var validTargetVariableValues = CheckVariablesForPossibleTargetVariables(dataset).Select(x => new StringValue(x).AsReadOnly()).ToList();
283      var target = validTargetVariableValues.Where(x => x.Value == targetVariable).DefaultIfEmpty(validTargetVariableValues.First()).First();
284
285      Parameters.Add(new ConstrainedValueParameter<StringValue>(TargetVariableParameterName, new ItemSet<StringValue>(validTargetVariableValues), target));
286      Parameters.Add(new FixedValueParameter<StringMatrix>(ClassNamesParameterName, ""));
287      Parameters.Add(new FixedValueParameter<DoubleMatrix>(ClassificationPenaltiesParameterName, ""));
288
289      ResetTargetVariableDependentMembers();
290      RegisterParameterEvents();
291    }
292
293    private static IEnumerable<string> CheckVariablesForPossibleTargetVariables(Dataset dataset) {
294      int maxSamples = Math.Min(InspectedRowsToDetermineTargets, dataset.Rows);
295      var validTargetVariables = (from v in dataset.DoubleVariables
296                                  let distinctValues = dataset.GetDoubleValues(v)
297                                    .Take(maxSamples)
298                                    .Distinct()
299                                    .Count()
300                                  where distinctValues < MaximumNumberOfClasses
301                                  select v).ToArray();
302
303      if (!validTargetVariables.Any())
304        throw new ArgumentException("Import of classification problem data was not successful, because no target variable was found." +
305          " A target variable must have at most " + MaximumNumberOfClasses + " distinct values to be applicable to classification.");
306      return validTargetVariables;
307    }
308
309
310    private void ResetTargetVariableDependentMembers() {
311      DeregisterParameterEvents();
312
313      classNames = null;
314      ((IStringConvertibleMatrix)ClassNamesParameter.Value).Columns = 1;
315      ((IStringConvertibleMatrix)ClassNamesParameter.Value).Rows = ClassValues.Count;
316      for (int i = 0; i < Classes; i++)
317        ClassNamesParameter.Value[i, 0] = "Class " + ClassValues[i];
318      ClassNamesParameter.Value.ColumnNames = new List<string>() { "ClassNames" };
319      ClassNamesParameter.Value.RowNames = ClassValues.Select(s => "ClassValue: " + s);
320
321      classificationPenaltiesCache.Clear();
322      ((ValueParameter<DoubleMatrix>)ClassificationPenaltiesParameter).ReactOnValueToStringChangedAndValueItemImageChanged = false;
323      ((IStringConvertibleMatrix)ClassificationPenaltiesParameter.Value).Rows = Classes;
324      ((IStringConvertibleMatrix)ClassificationPenaltiesParameter.Value).Columns = Classes;
325      ClassificationPenaltiesParameter.Value.RowNames = ClassNames.Select(name => "Actual " + name);
326      ClassificationPenaltiesParameter.Value.ColumnNames = ClassNames.Select(name => "Estimated " + name);
327      for (int i = 0; i < Classes; i++) {
328        for (int j = 0; j < Classes; j++) {
329          if (i != j) ClassificationPenaltiesParameter.Value[i, j] = 1;
330          else ClassificationPenaltiesParameter.Value[i, j] = 0;
331        }
332      }
333      ((ValueParameter<DoubleMatrix>)ClassificationPenaltiesParameter).ReactOnValueToStringChangedAndValueItemImageChanged = true;
334      RegisterParameterEvents();
335    }
336
337    public string GetClassName(double classValue) {
338      if (!ClassValues.Contains(classValue)) throw new ArgumentException();
339      int index = ClassValues.IndexOf(classValue);
340      return ClassNames[index];
341    }
342    public double GetClassValue(string className) {
343      if (!ClassNames.Contains(className)) throw new ArgumentException();
344      int index = ClassNames.IndexOf(className);
345      return ClassValues[index];
346    }
347    public void SetClassName(double classValue, string className) {
348      if (!classValues.Contains(classValue)) throw new ArgumentException();
349      int index = ClassValues.IndexOf(classValue);
350      ClassNames[index] = className;
351      ClassNamesParameter.Value[index, 0] = className;
352    }
353
354    public double GetClassificationPenalty(string correctClassName, string estimatedClassName) {
355      return GetClassificationPenalty(GetClassValue(correctClassName), GetClassValue(estimatedClassName));
356    }
357    public double GetClassificationPenalty(double correctClassValue, double estimatedClassValue) {
358      var key = Tuple.Create(correctClassValue, estimatedClassValue);
359      if (!classificationPenaltiesCache.ContainsKey(key)) {
360        int correctClassIndex = ClassValues.IndexOf(correctClassValue);
361        int estimatedClassIndex = ClassValues.IndexOf(estimatedClassValue);
362        classificationPenaltiesCache[key] = ClassificationPenaltiesParameter.Value[correctClassIndex, estimatedClassIndex];
363      }
364      return classificationPenaltiesCache[key];
365    }
366    public void SetClassificationPenalty(string correctClassName, string estimatedClassName, double penalty) {
367      SetClassificationPenalty(GetClassValue(correctClassName), GetClassValue(estimatedClassName), penalty);
368    }
369    public void SetClassificationPenalty(double correctClassValue, double estimatedClassValue, double penalty) {
370      var key = Tuple.Create(correctClassValue, estimatedClassValue);
371      int correctClassIndex = ClassValues.IndexOf(correctClassValue);
372      int estimatedClassIndex = ClassValues.IndexOf(estimatedClassValue);
373
374      ClassificationPenaltiesParameter.Value[correctClassIndex, estimatedClassIndex] = penalty;
375    }
376
377    #region events
378    private void RegisterParameterEvents() {
379      TargetVariableParameter.ValueChanged += new EventHandler(TargetVariableParameter_ValueChanged);
380      ClassNamesParameter.Value.Reset += new EventHandler(Parameter_ValueChanged);
381      ClassNamesParameter.Value.ItemChanged += new EventHandler<EventArgs<int, int>>(MatrixParameter_ItemChanged);
382      ClassificationPenaltiesParameter.Value.Reset += new EventHandler(Parameter_ValueChanged);
383      ClassificationPenaltiesParameter.Value.ItemChanged += new EventHandler<EventArgs<int, int>>(MatrixParameter_ItemChanged);
384    }
385    private void DeregisterParameterEvents() {
386      TargetVariableParameter.ValueChanged -= new EventHandler(TargetVariableParameter_ValueChanged);
387      ClassNamesParameter.Value.Reset -= new EventHandler(Parameter_ValueChanged);
388      ClassNamesParameter.Value.ItemChanged -= new EventHandler<EventArgs<int, int>>(MatrixParameter_ItemChanged);
389      ClassificationPenaltiesParameter.Value.Reset -= new EventHandler(Parameter_ValueChanged);
390      ClassificationPenaltiesParameter.Value.ItemChanged -= new EventHandler<EventArgs<int, int>>(MatrixParameter_ItemChanged);
391    }
392
393    private void TargetVariableParameter_ValueChanged(object sender, EventArgs e) {
394      classValues = null;
395      ResetTargetVariableDependentMembers();
396      OnChanged();
397    }
398    private void Parameter_ValueChanged(object sender, EventArgs e) {
399      OnChanged();
400    }
401    private void MatrixParameter_ItemChanged(object sender, EventArgs<int, int> e) {
402      OnChanged();
403    }
404    #endregion
405
406    #region Import from file
407    public static ClassificationProblemData ImportFromFile(string fileName) {
408      TableFileParser csvFileParser = new TableFileParser();
409      csvFileParser.Parse(fileName);
410
411      Dataset dataset = new Dataset(csvFileParser.VariableNames, csvFileParser.Values);
412      dataset.Name = Path.GetFileName(fileName);
413
414      ClassificationProblemData problemData = new ClassificationProblemData(dataset, dataset.DoubleVariables.Skip(1), dataset.DoubleVariables.First());
415      problemData.Name = "Data imported from " + Path.GetFileName(fileName);
416      return problemData;
417    }
418    #endregion
419  }
420}
Note: See TracBrowser for help on using the repository browser.