Changeset 17835

Timestamp:

02/12/21 16:26:36 (4 years ago)

Author:

bburlacu

Message:

#3102: Add ClassificationProblemData constructor that explicitly takes class names and positive class value arguments, adapt code.

Location:

trunk

Files:

• HeuristicLab.Algorithms.DataAnalysis/3.4/CrossValidation.cs (modified) (2 diffs)
• HeuristicLab.Problems.DataAnalysis.Symbolic.Classification.Views/3.4/SolutionComparisonView.cs (modified) (1 diff)
• HeuristicLab.Problems.DataAnalysis/3.4/Implementation/Classification/ClassificationEnsembleProblemData.cs (modified) (2 diffs)
• HeuristicLab.Problems.DataAnalysis/3.4/Implementation/Classification/ClassificationEnsembleSolution.cs (modified) (1 diff)
• HeuristicLab.Problems.DataAnalysis/3.4/Implementation/Classification/ClassificationProblemData.cs (modified) (3 diffs)

trunk/HeuristicLab.Algorithms.DataAnalysis/3.4/CrossValidation.cs

@@ -521 +521 @@
       foreach (KeyValuePair<string, List<IClassificationSolution>> solutions in resultSolutions) {
         // at least one algorithm (GBT with logistic regression loss) produces a classification solution even though the original problem is a regression problem.
-        var targetVariable = solutions.Value.First().ProblemData.TargetVariable;
         var dataset = (Dataset)Problem.ProblemData.Dataset;
         if (ShuffleSamples.Value) {
@@ -527 +526 @@
           dataset = dataset.Shuffle(random);
         }
-        var problemDataClone = new ClassificationProblemData(dataset, Problem.ProblemData.AllowedInputVariables, targetVariable);
+        var problemData = (IClassificationProblemData)Problem.ProblemData;
+        var problemDataClone = new ClassificationProblemData(dataset, problemData.AllowedInputVariables, problemData.TargetVariable, problemData.ClassNames, problemData.PositiveClass);
         // set partitions of problem data clone correctly
         problemDataClone.TrainingPartition.Start = SamplesStart.Value; problemDataClone.TrainingPartition.End = SamplesEnd.Value;

trunk/HeuristicLab.Problems.DataAnalysis.Symbolic.Classification.Views/3.4/SolutionComparisonView.cs

@@ -90 +90 @@
 
       var newDs = new Dataset(variableNames, variableValues);
-      var newProblemData = new ClassificationProblemData(newDs, variableNames.Take(variableNames.Length - 1), variableNames.Last());
-
-      foreach (var classValue in problemData.ClassValues) {
-        newProblemData.SetClassName(classValue, problemData.GetClassName(classValue));
-      }
-      newProblemData.PositiveClass = problemData.PositiveClass;
+      var newProblemData = new ClassificationProblemData(newDs, variableNames.Take(variableNames.Length - 1), variableNames.Last(), problemData.ClassNames, problemData.PositiveClass);
 
       newProblemData.TrainingPartition.Start = problemData.TrainingPartition.Start;

trunk/HeuristicLab.Problems.DataAnalysis/3.4/Implementation/Classification/ClassificationEnsembleProblemData.cs

@@ -73 +73 @@
 
     public ClassificationEnsembleProblemData() : base() { }
+
     public ClassificationEnsembleProblemData(IClassificationProblemData classificationProblemData)
-      : base(classificationProblemData.Dataset, classificationProblemData.AllowedInputVariables, classificationProblemData.TargetVariable) {
-      this.TrainingPartition.Start = classificationProblemData.TrainingPartition.Start;
-      this.TrainingPartition.End = classificationProblemData.TrainingPartition.End;
-      this.TestPartition.Start = classificationProblemData.TestPartition.Start;
-      this.TestPartition.End = classificationProblemData.TestPartition.End;
-      this.PositiveClass = classificationProblemData.PositiveClass;
+      : base(classificationProblemData) {
     }
 
@@ -85 +81 @@
       : base(dataset, allowedInputVariables, targetVariable) {
     }
+
+    public ClassificationEnsembleProblemData(Dataset dataset, IEnumerable<string> allowedInputVariables, string targetVariable, IEnumerable<string> classNames, string positiveClass = null)
+      : base(dataset, allowedInputVariables, targetVariable, classNames, positiveClass) {
+    }
   }
 }

trunk/HeuristicLab.Problems.DataAnalysis/3.4/Implementation/Classification/ClassificationEnsembleSolution.cs

@@ -260 +260 @@
       evaluationCache.Clear();
 
-      IClassificationProblemData problemData = new ClassificationProblemData(ProblemData.Dataset,
-                                                                     ProblemData.AllowedInputVariables,
-                                                                     ProblemData.TargetVariable);
-      problemData.TrainingPartition.Start = ProblemData.TrainingPartition.Start;
-      problemData.TrainingPartition.End = ProblemData.TrainingPartition.End;
-      problemData.TestPartition.Start = ProblemData.TestPartition.Start;
-      problemData.TestPartition.End = ProblemData.TestPartition.End;
+      IClassificationProblemData problemData = new ClassificationProblemData(ProblemData);
 
       foreach (var solution in ClassificationSolutions) {

trunk/HeuristicLab.Problems.DataAnalysis/3.4/Implementation/Classification/ClassificationProblemData.cs

@@ -311 +311 @@
     }
 
-    public ClassificationProblemData() : this(defaultDataset, defaultAllowedInputVariables, defaultTargetVariable) { }
+    public ClassificationProblemData() : this(defaultDataset, defaultAllowedInputVariables, defaultTargetVariable, Enumerable.Empty<string>()) { }
 
     public ClassificationProblemData(IClassificationProblemData classificationProblemData)
-      : this(classificationProblemData.Dataset, classificationProblemData.AllowedInputVariables, classificationProblemData.TargetVariable) {
+      : this(classificationProblemData.Dataset, classificationProblemData.AllowedInputVariables, classificationProblemData.TargetVariable, classificationProblemData.ClassNames, classificationProblemData.PositiveClass) {
+      
       TrainingPartition.Start = classificationProblemData.TrainingPartition.Start;
       TrainingPartition.End = classificationProblemData.TrainingPartition.End;
       TestPartition.Start = classificationProblemData.TestPartition.Start;
       TestPartition.End = classificationProblemData.TestPartition.End;
-
-      for (int i = 0; i < classificationProblemData.ClassNames.Count(); i++)
-        ClassNamesParameter.Value[i, 0] = classificationProblemData.ClassNames.ElementAt(i);
-
-      //mkommend: The positive class depends on the class names and as a result must only be set after the classe names parameter.
-      PositiveClass = classificationProblemData.PositiveClass;
-
+      
       for (int i = 0; i < Classes; i++) {
         for (int j = 0; j < Classes; j++) {
@@ -334 +329 @@
 
     public ClassificationProblemData(IDataset dataset, IEnumerable<string> allowedInputVariables, string targetVariable, IEnumerable<ITransformation> transformations = null)
+      : this(dataset, allowedInputVariables, targetVariable, Enumerable.Empty<string>(), null, transformations) { }
+
+    public ClassificationProblemData(IDataset dataset, IEnumerable<string> allowedInputVariables, string targetVariable,
+      IEnumerable<string> classNames,
+      string positiveClass = null, // can be null in which case it's set as the first class name
+      IEnumerable<ITransformation> transformations = null)
       : base(dataset, allowedInputVariables, transformations ?? Enumerable.Empty<ITransformation>()) {
       var validTargetVariableValues = CheckVariablesForPossibleTargetVariables(dataset).Select(x => new StringValue(x).AsReadOnly()).ToList();
@@ -339 +340 @@
 
       Parameters.Add(new ConstrainedValueParameter<StringValue>(TargetVariableParameterName, new ItemSet<StringValue>(validTargetVariableValues), target));
-      Parameters.Add(new FixedValueParameter<StringMatrix>(ClassNamesParameterName, ""));
+      Parameters.Add(new FixedValueParameter<StringMatrix>(ClassNamesParameterName, "", new StringMatrix()));
       Parameters.Add(new ConstrainedValueParameter<StringValue>(PositiveClassParameterName, "The positive class which is used for quality measure calculation (e.g., specifity, sensitivity,...)"));
       Parameters.Add(new FixedValueParameter<DoubleMatrix>(ClassificationPenaltiesParameterName, ""));
 
       RegisterParameterEvents();
-      ResetTargetVariableDependentMembers();
+      ResetTargetVariableDependentMembers(); // correctly set the values of the parameters added above
+
+      // set the class names
+      if (classNames.Any()) {
+        // better to allocate lists because we use these multiple times below
+        var names = classNames.ToList();
+        var values = ClassValues.ToList();
+
+        if (names.Count != values.Count) {
+          throw new ArgumentException();
+        }
+
+        ((IStringConvertibleMatrix)ClassNamesParameter.Value).Columns = 1;
+        ((IStringConvertibleMatrix)ClassNamesParameter.Value).Rows = names.Count;
+
+        for (int i = 0; i < names.Count; ++i) {
+          SetClassName(values[i], names[i]);
+        }
+      }
+
+      // set the positive class value
+      if (positiveClass != null) {
+        PositiveClass = positiveClass;
+      }
     }