#region License Information
/* HeuristicLab
* Copyright (C) 2002-2014 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
*
* This file is part of HeuristicLab.
*
* HeuristicLab is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
*
* HeuristicLab is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with HeuristicLab. If not, see .
*/
#endregion
using System.Collections.Generic;
using System.Linq;
using HeuristicLab.Common;
using HeuristicLab.Data;
using HeuristicLab.Optimization;
using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
namespace HeuristicLab.Problems.DataAnalysis {
[StorableClass]
public abstract class ClassificationSolutionBase : DataAnalysisSolution, IClassificationSolution {
private const string TrainingAccuracyResultName = "Accuracy (training)";
private const string TestAccuracyResultName = "Accuracy (test)";
private const string TrainingNormalizedGiniCoefficientResultName = "Normalized Gini Coefficient (training)";
private const string TestNormalizedGiniCoefficientResultName = "Normalized Gini Coefficient (test)";
private const string TrainingTruePositiveRateResultName = "True positive rate (training)";
private const string TrainingTrueNegativeRateResultName = "True negative rate (training)";
private const string TrainingPositivePredictiveValueResultName = "Positive predictive value (training)";
private const string TrainingNegativePredictiveValueResultName = "Negative predictive value (training)";
private const string TrainingFalsePositiveRateResultName = "False positive rate (training)";
private const string TrainingFalseDiscoveryRateResultName = "False discovery rate (training)";
private const string TestTruePositiveRateResultName = "True positive rate (test)";
private const string TestTrueNegativeRateResultName = "True negative rate (test)";
private const string TestPositivePredictiveValueResultName = "Positive predictive value (test)";
private const string TestNegativePredictiveValueResultName = "Negative predictive value (test)";
private const string TestFalsePositiveRateResultName = "False positive rate (test)";
private const string TestFalseDiscoveryRateResultName = "False discovery rate (test)";
private const string QualityMeasuresResultName = "Classification Quality Measures";
public new IClassificationModel Model {
get { return (IClassificationModel)base.Model; }
protected set { base.Model = value; }
}
public new IClassificationProblemData ProblemData {
get { return (IClassificationProblemData)base.ProblemData; }
set { base.ProblemData = value; }
}
#region Results
public double TrainingAccuracy {
get { return ((DoubleValue)this[TrainingAccuracyResultName].Value).Value; }
private set { ((DoubleValue)this[TrainingAccuracyResultName].Value).Value = value; }
}
public double TestAccuracy {
get { return ((DoubleValue)this[TestAccuracyResultName].Value).Value; }
private set { ((DoubleValue)this[TestAccuracyResultName].Value).Value = value; }
}
public double TrainingNormalizedGiniCoefficient {
get { return ((DoubleValue)this[TrainingNormalizedGiniCoefficientResultName].Value).Value; }
protected set { ((DoubleValue)this[TrainingNormalizedGiniCoefficientResultName].Value).Value = value; }
}
public double TestNormalizedGiniCoefficient {
get { return ((DoubleValue)this[TestNormalizedGiniCoefficientResultName].Value).Value; }
protected set { ((DoubleValue)this[TestNormalizedGiniCoefficientResultName].Value).Value = value; }
}
#region Quality Measures
public ResultCollection QualityMeasures {
get { return ((ResultCollection)this[QualityMeasuresResultName].Value); }
protected set { (this[QualityMeasuresResultName].Value) = value; }
}
public double TrainingTruePositiveRate {
get { return ((DoubleValue)QualityMeasures[TrainingTruePositiveRateResultName].Value).Value; }
protected set { ((DoubleValue)QualityMeasures[TrainingTruePositiveRateResultName].Value).Value = value; }
}
public double TrainingTrueNegativeRate {
get { return ((DoubleValue)QualityMeasures[TrainingTrueNegativeRateResultName].Value).Value; }
protected set { ((DoubleValue)QualityMeasures[TrainingTrueNegativeRateResultName].Value).Value = value; }
}
public double TrainingPositivePredictiveValue {
get { return ((DoubleValue)QualityMeasures[TrainingPositivePredictiveValueResultName].Value).Value; }
protected set { ((DoubleValue)QualityMeasures[TrainingPositivePredictiveValueResultName].Value).Value = value; }
}
public double TrainingNegativePredictiveValue {
get { return ((DoubleValue)QualityMeasures[TrainingNegativePredictiveValueResultName].Value).Value; }
protected set { ((DoubleValue)QualityMeasures[TrainingNegativePredictiveValueResultName].Value).Value = value; }
}
public double TrainingFalsePositiveRate {
get { return ((DoubleValue)QualityMeasures[TrainingFalsePositiveRateResultName].Value).Value; }
protected set { ((DoubleValue)QualityMeasures[TrainingFalsePositiveRateResultName].Value).Value = value; }
}
public double TrainingFalseDiscoveryRate {
get { return ((DoubleValue)QualityMeasures[TrainingFalseDiscoveryRateResultName].Value).Value; }
protected set { ((DoubleValue)QualityMeasures[TrainingFalseDiscoveryRateResultName].Value).Value = value; }
}
public double TestTruePositiveRate {
get { return ((DoubleValue)QualityMeasures[TestTruePositiveRateResultName].Value).Value; }
protected set { ((DoubleValue)QualityMeasures[TestTruePositiveRateResultName].Value).Value = value; }
}
public double TestTrueNegativeRate {
get { return ((DoubleValue)QualityMeasures[TestTrueNegativeRateResultName].Value).Value; }
protected set { ((DoubleValue)QualityMeasures[TestTrueNegativeRateResultName].Value).Value = value; }
}
public double TestPositivePredictiveValue {
get { return ((DoubleValue)QualityMeasures[TestPositivePredictiveValueResultName].Value).Value; }
protected set { ((DoubleValue)QualityMeasures[TestPositivePredictiveValueResultName].Value).Value = value; }
}
public double TestNegativePredictiveValue {
get { return ((DoubleValue)QualityMeasures[TestNegativePredictiveValueResultName].Value).Value; }
protected set { ((DoubleValue)QualityMeasures[TestNegativePredictiveValueResultName].Value).Value = value; }
}
public double TestFalsePositiveRate {
get { return ((DoubleValue)QualityMeasures[TestFalsePositiveRateResultName].Value).Value; }
protected set { ((DoubleValue)QualityMeasures[TestFalsePositiveRateResultName].Value).Value = value; }
}
public double TestFalseDiscoveryRate {
get { return ((DoubleValue)QualityMeasures[TestFalseDiscoveryRateResultName].Value).Value; }
protected set { ((DoubleValue)QualityMeasures[TestFalseDiscoveryRateResultName].Value).Value = value; }
}
#endregion
#endregion
[StorableConstructor]
protected ClassificationSolutionBase(bool deserializing) : base(deserializing) { }
protected ClassificationSolutionBase(ClassificationSolutionBase original, Cloner cloner)
: base(original, cloner) {
}
protected ClassificationSolutionBase(IClassificationModel model, IClassificationProblemData problemData)
: base(model, problemData) {
Add(new Result(TrainingAccuracyResultName, "Accuracy of the model on the training partition (percentage of correctly classified instances).", new PercentValue()));
Add(new Result(TestAccuracyResultName, "Accuracy of the model on the test partition (percentage of correctly classified instances).", new PercentValue()));
Add(new Result(TrainingNormalizedGiniCoefficientResultName, "Normalized Gini coefficient of the model on the training partition.", new DoubleValue()));
Add(new Result(TestNormalizedGiniCoefficientResultName, "Normalized Gini coefficient of the model on the test partition.", new DoubleValue()));
AddQualityMeasuresResultCollection();
}
[StorableHook(HookType.AfterDeserialization)]
private void AfterDeserialization() {
if (!this.ContainsKey(TrainingNormalizedGiniCoefficientResultName))
Add(new Result(TrainingNormalizedGiniCoefficientResultName, "Normalized Gini coefficient of the model on the training partition.", new DoubleValue()));
if (!this.ContainsKey(TestNormalizedGiniCoefficientResultName))
Add(new Result(TestNormalizedGiniCoefficientResultName, "Normalized Gini coefficient of the model on the test partition.", new DoubleValue()));
if (!this.ContainsKey(QualityMeasuresResultName))
AddQualityMeasuresResultCollection();
}
protected void AddQualityMeasuresResultCollection() {
ResultCollection qualityMeasuresResult = new ResultCollection();
qualityMeasuresResult.Add(new Result(TrainingTruePositiveRateResultName, "Sensitivity/True positive rate of the model on the training partition\n(TP/(TP+FN)).", new PercentValue()));
qualityMeasuresResult.Add(new Result(TrainingTrueNegativeRateResultName, "Specificity/True negative rate of the model on the training partition\n(TN/(FP+TN)).", new PercentValue()));
qualityMeasuresResult.Add(new Result(TrainingPositivePredictiveValueResultName, "Precision/Positive predictive value of the model on the training partition\n(TP/(TP+FP)).", new PercentValue()));
qualityMeasuresResult.Add(new Result(TrainingNegativePredictiveValueResultName, "Negative predictive value of the model on the training partition\n(TN/(TN+FN)).", new PercentValue()));
qualityMeasuresResult.Add(new Result(TrainingFalsePositiveRateResultName, "The false positive rate is the complement of the true negative rate of the model on the training partition.", new PercentValue()));
qualityMeasuresResult.Add(new Result(TrainingFalseDiscoveryRateResultName, "The false discovery rate is the complement of the positive predictive value of the model on the training partition.", new PercentValue()));
qualityMeasuresResult.Add(new Result(TestTruePositiveRateResultName, "Sensitivity/True positive rate of the model on the test partition\n(TP/(TP+FN)).", new PercentValue()));
qualityMeasuresResult.Add(new Result(TestTrueNegativeRateResultName, "Specificity/True negative rate of the model on the test partition\n(TN/(FP+TN)).", new PercentValue()));
qualityMeasuresResult.Add(new Result(TestPositivePredictiveValueResultName, "Precision/Positive predictive value of the model on the test partition\n(TP/(TP+FP)).", new PercentValue()));
qualityMeasuresResult.Add(new Result(TestNegativePredictiveValueResultName, "Negative predictive value of the model on the test partition\n(TN/(TN+FN)).", new PercentValue()));
qualityMeasuresResult.Add(new Result(TestFalsePositiveRateResultName, "The false positive rate is the complement of the true negative rate of the model on the test partition.", new PercentValue()));
qualityMeasuresResult.Add(new Result(TestFalseDiscoveryRateResultName, "The false discovery rate is the complement of the positive predictive value of the model on the test partition.", new PercentValue()));
Add(new Result(QualityMeasuresResultName, "Classification quality measures.\nIn Multiclass Classification all misclassifications of the negative class will be treated as true negatives.", qualityMeasuresResult));
}
protected void CalculateClassificationResults() {
double[] estimatedTrainingClassValues = EstimatedTrainingClassValues.ToArray(); // cache values
double[] originalTrainingClassValues = ProblemData.Dataset.GetDoubleValues(ProblemData.TargetVariable, ProblemData.TrainingIndices).ToArray();
double[] estimatedTestClassValues = EstimatedTestClassValues.ToArray(); // cache values
double[] originalTestClassValues = ProblemData.Dataset.GetDoubleValues(ProblemData.TargetVariable, ProblemData.TestIndices).ToArray();
var positiveClassName = ProblemData.PositiveClassName;
double positiveClassValue = ProblemData.GetClassValue(positiveClassName);
QualityCalculator trainingQualityCalculator = new QualityCalculator(positiveClassValue);
QualityCalculator testQualityCalculator = new QualityCalculator(positiveClassValue);
OnlineCalculatorError errorState;
double trainingAccuracy = OnlineAccuracyCalculator.Calculate(originalTrainingClassValues, estimatedTrainingClassValues, out errorState);
if (errorState != OnlineCalculatorError.None) trainingAccuracy = double.NaN;
double testAccuracy = OnlineAccuracyCalculator.Calculate(originalTestClassValues, estimatedTestClassValues, out errorState);
if (errorState != OnlineCalculatorError.None) testAccuracy = double.NaN;
TrainingAccuracy = trainingAccuracy;
TestAccuracy = testAccuracy;
double trainingNormalizedGini = NormalizedGiniCalculator.Calculate(originalTrainingClassValues, estimatedTrainingClassValues, out errorState);
if (errorState != OnlineCalculatorError.None) trainingNormalizedGini = double.NaN;
double testNormalizedGini = NormalizedGiniCalculator.Calculate(originalTestClassValues, estimatedTestClassValues, out errorState);
if (errorState != OnlineCalculatorError.None) testNormalizedGini = double.NaN;
TrainingNormalizedGiniCoefficient = trainingNormalizedGini;
TestNormalizedGiniCoefficient = testNormalizedGini;
//quality measures training partition
trainingQualityCalculator.Calculate(originalTrainingClassValues, estimatedTrainingClassValues, out errorState);
if (errorState != OnlineCalculatorError.None) {
TrainingTruePositiveRate = double.NaN;
TrainingTrueNegativeRate = double.NaN;
TrainingPositivePredictiveValue = double.NaN;
TrainingNegativePredictiveValue = double.NaN;
TrainingFalsePositiveRate = double.NaN;
TrainingFalseDiscoveryRate = double.NaN;
} else {
TrainingTruePositiveRate = trainingQualityCalculator.TruePositiveRate;
TrainingTrueNegativeRate = trainingQualityCalculator.TrueNegativeRate;
TrainingPositivePredictiveValue = trainingQualityCalculator.PositivePredictiveValue;
TrainingNegativePredictiveValue = trainingQualityCalculator.NegativePredictiveValue;
TrainingFalsePositiveRate = trainingQualityCalculator.FalsePositiveRate;
TrainingFalseDiscoveryRate = trainingQualityCalculator.FalseDiscoveryRate;
}
//quality measures test partition
testQualityCalculator.Calculate(originalTestClassValues, estimatedTestClassValues, out errorState);
if (errorState != OnlineCalculatorError.None) {
TestTruePositiveRate = double.NaN;
TestTrueNegativeRate = double.NaN;
TestPositivePredictiveValue = double.NaN;
TestNegativePredictiveValue = double.NaN;
TestFalsePositiveRate = double.NaN;
TestFalseDiscoveryRate = double.NaN;
} else {
TestTruePositiveRate = testQualityCalculator.TruePositiveRate;
TestTrueNegativeRate = testQualityCalculator.TrueNegativeRate;
TestPositivePredictiveValue = testQualityCalculator.PositivePredictiveValue;
TestNegativePredictiveValue = testQualityCalculator.NegativePredictiveValue;
TestFalsePositiveRate = testQualityCalculator.FalsePositiveRate;
TestFalseDiscoveryRate = testQualityCalculator.FalseDiscoveryRate;
}
}
public abstract IEnumerable EstimatedClassValues { get; }
public abstract IEnumerable EstimatedTrainingClassValues { get; }
public abstract IEnumerable EstimatedTestClassValues { get; }
public abstract IEnumerable GetEstimatedClassValues(IEnumerable rows);
protected override void RecalculateResults() {
CalculateClassificationResults();
}
}
}