Changeset 11031


Ignore:
Timestamp:
06/20/14 17:08:41 (5 years ago)
Author:
gkronber
Message:

#1758 made appropriate changes to the problem data classes for trading and time series prognosis

Location:
trunk/sources
Files:
4 edited

Legend:

Unmodified
Added
Removed
  • trunk/sources/HeuristicLab.Problems.DataAnalysis.Trading/3.4/ProblemData.cs

    r10540 r11031  
    16521652
    16531653    public override void AdjustProblemDataProperties(IDataAnalysisProblemData problemData) {
    1654       throw new NotImplementedException("TODO");
     1654      var data = problemData as ProblemData;
     1655      if (data == null) throw new ArgumentException("The problem data is not a problem data set for trading. Instead a " + problemData.GetType().GetPrettyName() + " was provided.", "problemData");
     1656
     1657      string errorMessage;
     1658      if (!data.IsProblemDataCompatible(this, out errorMessage)) {
     1659        throw new InvalidOperationException(errorMessage);
     1660      }
     1661
     1662      base.AdjustProblemDataProperties(data);
     1663
     1664      var toDelete = PriceChangeVariableParameter.ValidValues.ToList();
     1665      foreach (var entry in data.PriceChangeVariableParameter.ValidValues) {
     1666        if (toDelete.Any(x => x.Value == entry.Value)) {
     1667          toDelete.RemoveAll(x => x.Value == entry.Value);
     1668        } else {
     1669          PriceChangeVariableParameter.ValidValues.Add(new StringValue(entry.Value));
     1670        }
     1671      }
     1672      PriceChangeVariableParameter.Value =
     1673        PriceChangeVariableParameter.ValidValues.Single(v => v.Value == data.PriceChangeVariable);
     1674
     1675      foreach (var varToDelete in toDelete) PriceChangeVariableParameter.ValidValues.Remove(varToDelete);
     1676
     1677      TransactionCostsParameter.Value.Value = data.TransactionCosts;
     1678
     1679      OnChanged();
    16551680    }
    16561681  }
  • trunk/sources/HeuristicLab.Problems.DataAnalysis/3.4/Implementation/TimeSeriesPrognosis/TimeSeriesPrognosisProblemData.cs

    r10540 r11031  
    16211621    }
    16221622
     1623    protected override bool IsProblemDataCompatible(IDataAnalysisProblemData problemData, out string errorMessage) {
     1624      if (problemData == null) throw new ArgumentNullException("problemData", "The provided problemData is null.");
     1625      ITimeSeriesPrognosisProblemData timeseriesProblemData = problemData as ITimeSeriesPrognosisProblemData;
     1626      if (timeseriesProblemData == null)
     1627        throw new ArgumentException("The problem data is not a time-series problem data. Instead a " + problemData.GetType().GetPrettyName() + " was provided.", "problemData");
     1628
     1629      var returnValue = base.IsProblemDataCompatible(problemData, out errorMessage);
     1630      //check targetVariable
     1631      if (problemData.InputVariables.All(var => var.Value != TargetVariable)) {
     1632        errorMessage = string.Format("The target variable {0} is not present in the new problem data.", TargetVariable)
     1633                       + Environment.NewLine + errorMessage;
     1634        return false;
     1635      }
     1636      return returnValue;
     1637    }
     1638
    16231639    public override void AdjustProblemDataProperties(IDataAnalysisProblemData problemData) {
    16241640      TimeSeriesPrognosisProblemData timeSeriesProblemData = problemData as TimeSeriesPrognosisProblemData;
     
    16261642        throw new ArgumentException("The problem data is not a timeseries problem data. Instead a " + problemData.GetType().GetPrettyName() + " was provided.", "problemData");
    16271643
     1644      var trainingDataStart = TrainingIndices.First();
     1645     
    16281646      base.AdjustProblemDataProperties(problemData);
     1647     
     1648      TestPartition.Start = trainingDataStart;
    16291649
    16301650      TrainingHorizon = timeSeriesProblemData.TrainingHorizon;
  • trunk/sources/HeuristicLab.Problems.DataAnalysis/3.4/Implementation/TimeSeriesPrognosis/TimeSeriesPrognosisResults.cs

    r9456 r11031  
    369369      OnlineCalculatorError errorState;
    370370      var problemData = Solution.ProblemData;
     371      if (!problemData.TrainingIndices.Any()) return;
    371372      var model = Solution.Model;
    372373      //mean model
     
    415416      OnlineCalculatorError errorState;
    416417      var problemData = Solution.ProblemData;
     418      if (!problemData.TestIndices.Any()) return;
    417419      var model = Solution.Model;
    418       //mean model
    419       double trainingMean = problemData.Dataset.GetDoubleValues(problemData.TargetVariable, problemData.TrainingIndices).Average();
    420       var meanModel = new ConstantTimeSeriesPrognosisModel(trainingMean);
    421 
    422       //AR1 model
    423       double alpha, beta;
    424       IEnumerable<double> trainingStartValues = problemData.Dataset.GetDoubleValues(problemData.TargetVariable, problemData.TrainingIndices.Select(r => r - 1).Where(r => r > 0)).ToList();
    425       OnlineLinearScalingParameterCalculator.Calculate(problemData.Dataset.GetDoubleValues(problemData.TargetVariable, problemData.TrainingIndices.Where(x => x > 0)), trainingStartValues, out alpha, out beta, out errorState);
    426       var AR1model = new TimeSeriesPrognosisAutoRegressiveModel(problemData.TargetVariable, new double[] { beta }, alpha);
    427 
    428420      var testHorizions = problemData.TestIndices.Select(r => Math.Min(testHorizon, problemData.TestPartition.End - r)).ToList();
    429421      IEnumerable<IEnumerable<double>> testTargetValues = problemData.TestIndices.Zip(testHorizions, Enumerable.Range).Select(r => problemData.Dataset.GetDoubleValues(problemData.TargetVariable, r)).ToList();
    430422      IEnumerable<IEnumerable<double>> testEstimatedValues = model.GetPrognosedValues(problemData.Dataset, problemData.TestIndices, testHorizions).ToList();
    431423      IEnumerable<double> testStartValues = problemData.Dataset.GetDoubleValues(problemData.TargetVariable, problemData.TestIndices.Select(r => r - 1).Where(r => r > 0)).ToList();
    432       IEnumerable<IEnumerable<double>> testMeanModelPredictions = meanModel.GetPrognosedValues(problemData.Dataset, problemData.TestIndices, testHorizions).ToList();
    433       IEnumerable<IEnumerable<double>> testAR1ModelPredictions = AR1model.GetPrognosedValues(problemData.Dataset, problemData.TestIndices, testHorizions).ToList();
    434424
    435425      IEnumerable<double> originalTestValues = testTargetValues.SelectMany(x => x).ToList();
     
    453443      PrognosisTestWeightedDirectionalSymmetry = OnlineWeightedDirectionalSymmetryCalculator.Calculate(testStartValues, testTargetValues, testEstimatedValues, out errorState);
    454444      PrognosisTestWeightedDirectionalSymmetry = errorState == OnlineCalculatorError.None ? PrognosisTestWeightedDirectionalSymmetry : 0.0;
    455       PrognosisTestTheilsUStatisticAR1 = OnlineTheilsUStatisticCalculator.Calculate(testStartValues, testTargetValues, testAR1ModelPredictions, testEstimatedValues, out errorState);
    456       PrognosisTestTheilsUStatisticAR1 = errorState == OnlineCalculatorError.None ? PrognosisTestTheilsUStatisticAR1 : double.PositiveInfinity;
    457       PrognosisTestTheilsUStatisticMean = OnlineTheilsUStatisticCalculator.Calculate(testStartValues, testTargetValues, testMeanModelPredictions, testEstimatedValues, out errorState);
    458       PrognosisTestTheilsUStatisticMean = errorState == OnlineCalculatorError.None ? PrognosisTestTheilsUStatisticMean : double.PositiveInfinity;
     445
     446
     447      if (problemData.TrainingIndices.Any()) {
     448        //mean model
     449        double trainingMean = problemData.Dataset.GetDoubleValues(problemData.TargetVariable, problemData.TrainingIndices).Average();
     450        var meanModel = new ConstantTimeSeriesPrognosisModel(trainingMean);
     451
     452        //AR1 model
     453        double alpha, beta;
     454        IEnumerable<double> trainingStartValues = problemData.Dataset.GetDoubleValues(problemData.TargetVariable, problemData.TrainingIndices.Select(r => r - 1).Where(r => r > 0)).ToList();
     455        OnlineLinearScalingParameterCalculator.Calculate(problemData.Dataset.GetDoubleValues(problemData.TargetVariable, problemData.TrainingIndices.Where(x => x > 0)), trainingStartValues, out alpha, out beta, out errorState);
     456        var AR1model = new TimeSeriesPrognosisAutoRegressiveModel(problemData.TargetVariable, new double[] { beta }, alpha);
     457
     458        IEnumerable<IEnumerable<double>> testMeanModelPredictions = meanModel.GetPrognosedValues(problemData.Dataset, problemData.TestIndices, testHorizions).ToList();
     459        IEnumerable<IEnumerable<double>> testAR1ModelPredictions = AR1model.GetPrognosedValues(problemData.Dataset, problemData.TestIndices, testHorizions).ToList();
     460
     461        PrognosisTestTheilsUStatisticAR1 = OnlineTheilsUStatisticCalculator.Calculate(testStartValues, testTargetValues, testAR1ModelPredictions, testEstimatedValues, out errorState);
     462        PrognosisTestTheilsUStatisticAR1 = errorState == OnlineCalculatorError.None ? PrognosisTestTheilsUStatisticAR1 : double.PositiveInfinity;
     463        PrognosisTestTheilsUStatisticMean = OnlineTheilsUStatisticCalculator.Calculate(testStartValues, testTargetValues, testMeanModelPredictions, testEstimatedValues, out errorState);
     464        PrognosisTestTheilsUStatisticMean = errorState == OnlineCalculatorError.None ? PrognosisTestTheilsUStatisticMean : double.PositiveInfinity;
     465      }
    459466    }
    460467  }
  • trunk/sources/HeuristicLab.Problems.DataAnalysis/3.4/Implementation/TimeSeriesPrognosis/TimeSeriesPrognosisSolutionBase.cs

    r9462 r11031  
    149149    protected void CalculateTimeSeriesResults() {
    150150      OnlineCalculatorError errorState;
    151       double trainingMean = ProblemData.Dataset.GetDoubleValues(ProblemData.TargetVariable, ProblemData.TrainingIndices).Average();
     151      double trainingMean = ProblemData.TrainingIndices.Any() ? ProblemData.Dataset.GetDoubleValues(ProblemData.TargetVariable, ProblemData.TrainingIndices).Average() : double.NaN;
    152152      var meanModel = new ConstantTimeSeriesPrognosisModel(trainingMean);
    153153
     
    159159
    160160      #region Calculate training quality measures
    161       IEnumerable<double> trainingTargetValues = ProblemData.Dataset.GetDoubleValues(ProblemData.TargetVariable, ProblemData.TrainingIndices).ToList();
    162       IEnumerable<double> trainingEstimatedValues = EstimatedTrainingValues.ToList();
    163       IEnumerable<double> trainingMeanModelPredictions = meanModel.GetEstimatedValues(ProblemData.Dataset, ProblemData.TrainingIndices).ToList();
    164       IEnumerable<double> trainingAR1ModelPredictions = AR1model.GetEstimatedValues(ProblemData.Dataset, ProblemData.TrainingIndices).ToList();
    165 
    166       TrainingDirectionalSymmetry = OnlineDirectionalSymmetryCalculator.Calculate(trainingTargetValues.First(), trainingTargetValues, trainingEstimatedValues, out errorState);
    167       TrainingDirectionalSymmetry = errorState == OnlineCalculatorError.None ? TrainingDirectionalSymmetry : 0.0;
    168       TrainingWeightedDirectionalSymmetry = OnlineWeightedDirectionalSymmetryCalculator.Calculate(trainingTargetValues.First(), trainingTargetValues, trainingEstimatedValues, out errorState);
    169       TrainingWeightedDirectionalSymmetry = errorState == OnlineCalculatorError.None ? TrainingWeightedDirectionalSymmetry : 0.0;
    170       TrainingTheilsUStatisticAR1 = OnlineTheilsUStatisticCalculator.Calculate(trainingTargetValues.First(), trainingTargetValues, trainingAR1ModelPredictions, trainingEstimatedValues, out errorState);
    171       TrainingTheilsUStatisticAR1 = errorState == OnlineCalculatorError.None ? TrainingTheilsUStatisticAR1 : double.PositiveInfinity;
    172       TrainingTheilsUStatisticMean = OnlineTheilsUStatisticCalculator.Calculate(trainingTargetValues.First(), trainingTargetValues, trainingMeanModelPredictions, trainingEstimatedValues, out errorState);
    173       TrainingTheilsUStatisticMean = errorState == OnlineCalculatorError.None ? TrainingTheilsUStatisticMean : double.PositiveInfinity;
     161      if (ProblemData.TrainingIndices.Any()) {
     162        IEnumerable<double> trainingTargetValues = ProblemData.Dataset.GetDoubleValues(ProblemData.TargetVariable, ProblemData.TrainingIndices).ToList();
     163        IEnumerable<double> trainingEstimatedValues = EstimatedTrainingValues.ToList();
     164        IEnumerable<double> trainingMeanModelPredictions = meanModel.GetEstimatedValues(ProblemData.Dataset, ProblemData.TrainingIndices).ToList();
     165        IEnumerable<double> trainingAR1ModelPredictions = AR1model.GetEstimatedValues(ProblemData.Dataset, ProblemData.TrainingIndices).ToList();
     166
     167        TrainingDirectionalSymmetry = OnlineDirectionalSymmetryCalculator.Calculate(trainingTargetValues.First(), trainingTargetValues, trainingEstimatedValues, out errorState);
     168        TrainingDirectionalSymmetry = errorState == OnlineCalculatorError.None ? TrainingDirectionalSymmetry : 0.0;
     169        TrainingWeightedDirectionalSymmetry = OnlineWeightedDirectionalSymmetryCalculator.Calculate(trainingTargetValues.First(), trainingTargetValues, trainingEstimatedValues, out errorState);
     170        TrainingWeightedDirectionalSymmetry = errorState == OnlineCalculatorError.None ? TrainingWeightedDirectionalSymmetry : 0.0;
     171        TrainingTheilsUStatisticAR1 = OnlineTheilsUStatisticCalculator.Calculate(trainingTargetValues.First(), trainingTargetValues, trainingAR1ModelPredictions, trainingEstimatedValues, out errorState);
     172        TrainingTheilsUStatisticAR1 = errorState == OnlineCalculatorError.None ? TrainingTheilsUStatisticAR1 : double.PositiveInfinity;
     173        TrainingTheilsUStatisticMean = OnlineTheilsUStatisticCalculator.Calculate(trainingTargetValues.First(), trainingTargetValues, trainingMeanModelPredictions, trainingEstimatedValues, out errorState);
     174        TrainingTheilsUStatisticMean = errorState == OnlineCalculatorError.None ? TrainingTheilsUStatisticMean : double.PositiveInfinity;
     175      }
    174176      #endregion
    175177
    176178      #region Calculate test quality measures
    177       IEnumerable<double> testTargetValues = ProblemData.Dataset.GetDoubleValues(ProblemData.TargetVariable, ProblemData.TestIndices).ToList();
    178       IEnumerable<double> testEstimatedValues = EstimatedTestValues.ToList();
    179       IEnumerable<double> testMeanModelPredictions = meanModel.GetEstimatedValues(ProblemData.Dataset, ProblemData.TestIndices).ToList();
    180       IEnumerable<double> testAR1ModelPredictions = AR1model.GetEstimatedValues(ProblemData.Dataset, ProblemData.TestIndices).ToList();
    181 
    182       TestDirectionalSymmetry = OnlineDirectionalSymmetryCalculator.Calculate(testTargetValues.First(), testTargetValues, testEstimatedValues, out errorState);
    183       TestDirectionalSymmetry = errorState == OnlineCalculatorError.None ? TestDirectionalSymmetry : 0.0;
    184       TestWeightedDirectionalSymmetry = OnlineWeightedDirectionalSymmetryCalculator.Calculate(testTargetValues.First(), testTargetValues, testEstimatedValues, out errorState);
    185       TestWeightedDirectionalSymmetry = errorState == OnlineCalculatorError.None ? TestWeightedDirectionalSymmetry : 0.0;
    186       TestTheilsUStatisticAR1 = OnlineTheilsUStatisticCalculator.Calculate(testTargetValues.First(), testTargetValues, testAR1ModelPredictions, testEstimatedValues, out errorState);
    187       TestTheilsUStatisticAR1 = errorState == OnlineCalculatorError.None ? TestTheilsUStatisticAR1 : double.PositiveInfinity;
    188       TestTheilsUStatisticMean = OnlineTheilsUStatisticCalculator.Calculate(testTargetValues.First(), testTargetValues, testMeanModelPredictions, testEstimatedValues, out errorState);
    189       TestTheilsUStatisticMean = errorState == OnlineCalculatorError.None ? TestTheilsUStatisticMean : double.PositiveInfinity;
     179      if (ProblemData.TestIndices.Any()) {
     180        IEnumerable<double> testTargetValues = ProblemData.Dataset.GetDoubleValues(ProblemData.TargetVariable, ProblemData.TestIndices).ToList();
     181        IEnumerable<double> testEstimatedValues = EstimatedTestValues.ToList();
     182        IEnumerable<double> testMeanModelPredictions = meanModel.GetEstimatedValues(ProblemData.Dataset, ProblemData.TestIndices).ToList();
     183        IEnumerable<double> testAR1ModelPredictions = AR1model.GetEstimatedValues(ProblemData.Dataset, ProblemData.TestIndices).ToList();
     184
     185        TestDirectionalSymmetry = OnlineDirectionalSymmetryCalculator.Calculate(testTargetValues.First(), testTargetValues, testEstimatedValues, out errorState);
     186        TestDirectionalSymmetry = errorState == OnlineCalculatorError.None ? TestDirectionalSymmetry : 0.0;
     187        TestWeightedDirectionalSymmetry = OnlineWeightedDirectionalSymmetryCalculator.Calculate(testTargetValues.First(), testTargetValues, testEstimatedValues, out errorState);
     188        TestWeightedDirectionalSymmetry = errorState == OnlineCalculatorError.None ? TestWeightedDirectionalSymmetry : 0.0;
     189        TestTheilsUStatisticAR1 = OnlineTheilsUStatisticCalculator.Calculate(testTargetValues.First(), testTargetValues, testAR1ModelPredictions, testEstimatedValues, out errorState);
     190        TestTheilsUStatisticAR1 = errorState == OnlineCalculatorError.None ? TestTheilsUStatisticAR1 : double.PositiveInfinity;
     191        TestTheilsUStatisticMean = OnlineTheilsUStatisticCalculator.Calculate(testTargetValues.First(), testTargetValues, testMeanModelPredictions, testEstimatedValues, out errorState);
     192        TestTheilsUStatisticMean = errorState == OnlineCalculatorError.None ? TestTheilsUStatisticMean : double.PositiveInfinity;
     193      }
    190194      #endregion
    191195    }
Note: See TracChangeset for help on using the changeset viewer.