Changeset 14562 for branches/MemPRAlgorithm/HeuristicLab.Problems.DataAnalysis/3.4/Implementation

Timestamp:

01/13/17 12:55:06 (8 years ago)

Author:

abeham

Message:

#2701: Updated branch to trunk

Location:

branches/MemPRAlgorithm

Files:

• . (modified) (1 prop)
• HeuristicLab.Problems.DataAnalysis (modified) (1 prop)
• HeuristicLab.Problems.DataAnalysis/3.4/Implementation/DataAnalysisProblemData.cs (modified) (3 diffs)
• HeuristicLab.Problems.DataAnalysis/3.4/Implementation/Regression/RegressionSolutionVariableImpactsCalculator.cs (modified) (1 diff)
• HeuristicLab.Problems.DataAnalysis/3.4/Implementation/TimeSeriesPrognosis/TimeSeriesPrognosisSolution.cs (modified) (2 diffs)
• HeuristicLab.Problems.DataAnalysis/3.4/Implementation/TimeSeriesPrognosis/TimeSeriesPrognosisSolutionBase.cs (modified) (2 diffs)

branches/MemPRAlgorithm/HeuristicLab.Problems.DataAnalysis/3.4/Implementation/DataAnalysisProblemData.cs

@@ -41 +41 @@
 
     #region parameter properites
+    //mkommend: inserted parameter caching due to performance reasons
+    private IFixedValueParameter<Dataset> datasetParameter;
     public IFixedValueParameter<Dataset> DatasetParameter {
-      get { return (IFixedValueParameter<Dataset>)Parameters[DatasetParameterName]; }
-    }
+      get {
+        if (datasetParameter == null) datasetParameter = (IFixedValueParameter<Dataset>)Parameters[DatasetParameterName];
+        return datasetParameter;
+      }
+    }
+
+    private IFixedValueParameter<ReadOnlyCheckedItemList<StringValue>> inputVariablesParameter;
     public IFixedValueParameter<ReadOnlyCheckedItemList<StringValue>> InputVariablesParameter {
-      get { return (IFixedValueParameter<ReadOnlyCheckedItemList<StringValue>>)Parameters[InputVariablesParameterName]; }
-    }
+      get {
+        if (inputVariablesParameter == null) inputVariablesParameter = (IFixedValueParameter<ReadOnlyCheckedItemList<StringValue>>)Parameters[InputVariablesParameterName];
+        return inputVariablesParameter;
+      }
+    }
+
+    private IFixedValueParameter<IntRange> trainingPartitionParameter;
     public IFixedValueParameter<IntRange> TrainingPartitionParameter {
-      get { return (IFixedValueParameter<IntRange>)Parameters[TrainingPartitionParameterName]; }
-    }
+      get {
+        if (trainingPartitionParameter == null) trainingPartitionParameter = (IFixedValueParameter<IntRange>)Parameters[TrainingPartitionParameterName];
+        return trainingPartitionParameter;
+      }
+    }
+
+    private IFixedValueParameter<IntRange> testPartitionParameter;
     public IFixedValueParameter<IntRange> TestPartitionParameter {
-      get { return (IFixedValueParameter<IntRange>)Parameters[TestPartitionParameterName]; }
-    }
+      get {
+        if (testPartitionParameter == null) testPartitionParameter = (IFixedValueParameter<IntRange>)Parameters[TestPartitionParameterName];
+        return testPartitionParameter;
+      }
+    }
+
     public IFixedValueParameter<ReadOnlyItemList<ITransformation>> TransformationsParameter {
       get { return (IFixedValueParameter<ReadOnlyItemList<ITransformation>>)Parameters[TransformationsParameterName]; }
@@ -102 +123 @@
     public virtual bool IsTrainingSample(int index) {
       return index >= 0 && index < Dataset.Rows &&
-        TrainingPartition.Start <= index && index < TrainingPartition.End &&
-        (index < TestPartition.Start || TestPartition.End <= index);
+             TrainingPartition.Start <= index && index < TrainingPartition.End &&
+             (index < TestPartition.Start || TestPartition.End <= index);
     }
 
@@ -214 +235 @@
         InputVariables.SetItemCheckedState(inputVariable, variable != null && data.InputVariables.ItemChecked(variable));
       }
-
-      TrainingPartition.Start = TrainingPartition.End = 0;
-      TestPartition.Start = 0;
-      TestPartition.End = Dataset.Rows;
     }
   }

branches/MemPRAlgorithm/HeuristicLab.Problems.DataAnalysis/3.4/Implementation/Regression/RegressionSolutionVariableImpactsCalculator.cs

@@ -122 +122 @@
       }
 
-
       var impacts = new Dictionary<string, double>();
       var modifiableDataset = ((Dataset)dataset).ToModifiable();
 
-      foreach (var inputVariable in problemData.AllowedInputVariables) {
+      var inputvariables = new HashSet<string>(problemData.AllowedInputVariables.Union(solution.Model.VariablesUsedForPrediction));
+      var allowedInputVariables = dataset.VariableNames.Where(v => inputvariables.Contains(v)).ToList();
+
+      foreach (var inputVariable in allowedInputVariables) {
         var newEstimates = EvaluateModelWithReplacedVariable(solution.Model, inputVariable, modifiableDataset, rows, replacement);
         var newR2 = OnlinePearsonsRCalculator.Calculate(targetValues, newEstimates, out error);

branches/MemPRAlgorithm/HeuristicLab.Problems.DataAnalysis/3.4/Implementation/TimeSeriesPrognosis/TimeSeriesPrognosisSolution.cs

@@ -21 +21 @@
 
 using System.Collections.Generic;
+using System.Linq;
 using HeuristicLab.Common;
 using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
@@ -43 +44 @@
       return Model.GetPrognosedValues(ProblemData.Dataset, rows, horizons);
     }
+
+    public override IEnumerable<double> PrognosedTestValues {
+      get {
+        return Model.GetPrognosedValues(ProblemData.Dataset, ProblemData.TestIndices.Take(1),
+          new int[] { ProblemData.TestIndices.Count() }).First();
+      }
+    }
   }
 }

branches/MemPRAlgorithm/HeuristicLab.Problems.DataAnalysis/3.4/Implementation/TimeSeriesPrognosis/TimeSeriesPrognosisSolutionBase.cs

@@ -64 +64 @@
     }
 
+    public abstract IEnumerable<double> PrognosedTestValues { get; }
     public abstract IEnumerable<IEnumerable<double>> GetPrognosedValues(IEnumerable<int> rows, IEnumerable<int> horizon);
 
@@ -150 +151 @@
       OnlineCalculatorError errorState;
       double trainingMean = ProblemData.TrainingIndices.Any() ? ProblemData.Dataset.GetDoubleValues(ProblemData.TargetVariable, ProblemData.TrainingIndices).Average() : double.NaN;
-      var meanModel = new ConstantModel(trainingMean,ProblemData.TargetVariable);
+      var meanModel = new ConstantModel(trainingMean, ProblemData.TargetVariable);
 
       double alpha, beta;