Changeset 8921 for trunk/sources/HeuristicLab.Problems.DataAnalysis
 Timestamp:
 11/19/12 15:11:44 (11 years ago)
 File:

 1 edited
Legend:
 Unmodified
 Added
 Removed

trunk/sources/HeuristicLab.Problems.DataAnalysis/3.4/Implementation/Classification/ThresholdCalculators/NormalDistributionCutPointsThresholdCalculator.cs
r8917 r8921 71 71 } 72 72 } 73 73 74 double[] originalClasses = classMean.Keys.OrderBy(x => x).ToArray(); 74 75 int nClasses = originalClasses.Length; … … 98 99 thresholdList.Add(double.PositiveInfinity); 99 100 100 // determine class values for each partition separated by a threshold by calculating the density of all class distributions 101 // all points in the partition are classified as the class with the maximal density in the parition 102 if (thresholdList.Count == 2) { 101 102 // find the most likely class for the points between thresholds m 103 List<double> filteredThresholds = new List<double>(); 104 List<double> filteredClassValues = new List<double>(); 105 for (int i = 0; i < thresholdList.Count  1; i++) { 106 // determine class with maximal density mass between the thresholds 107 double maxDensity = DensityMass(thresholdList[i], thresholdList[i + 1], classMean[originalClasses[0]], classStdDev[originalClasses[0]]); 108 double maxDensityClassValue = originalClasses[0]; 109 foreach (var classValue in originalClasses.Skip(1)) { 110 double density = DensityMass(thresholdList[i], thresholdList[i + 1], classMean[classValue], classStdDev[classValue]); 111 if (density > maxDensity) { 112 maxDensity = density; 113 maxDensityClassValue = classValue; 114 } 115 } 116 if (maxDensity > double.NegativeInfinity && 117 (filteredClassValues.Count == 0  !maxDensityClassValue.IsAlmost(filteredClassValues.Last()))) { 118 filteredThresholds.Add(thresholdList[i]); 119 filteredClassValues.Add(maxDensityClassValue); 120 } 121 } 122 123 if (filteredThresholds.Count == 0  !double.IsNegativeInfinity(filteredThresholds.First())) { 103 124 // this happens if there are no thresholds (distributions for all classes are exactly the same) 125 // or when the CDF up to the first threshold is zero 104 126 // > all samples should be classified as the class with the most observations 105 127 // group observations by target class and select the class with largest count … … 107 129 .OrderBy(g => g.Count()) 108 130 .Last().Key; 109 thresholds = new double[] { double.NegativeInfinity }; 110 classValues = new double[] { mostFrequentClass }; 111 } else { 112 113 // at least one reasonable threshold ... 114 // find the most likely class for the points between thresholds m 115 List<double> filteredThresholds = new List<double>(); 116 List<double> filteredClassValues = new List<double>(); 117 for (int i = 0; i < thresholdList.Count  1; i++) { 118 // determine class with maximal density mass between the thresholds 119 double maxDensity = DensityMass(thresholdList[i], thresholdList[i + 1], classMean[originalClasses[0]], classStdDev[originalClasses[0]]); 120 double maxDensityClassValue = originalClasses[0]; 121 foreach (var classValue in originalClasses.Skip(1)) { 122 double density = DensityMass(thresholdList[i], thresholdList[i + 1], classMean[classValue], classStdDev[classValue]); 123 if (density > maxDensity) { 124 maxDensity = density; 125 maxDensityClassValue = classValue; 126 } 127 } 128 if (maxDensity > double.NegativeInfinity && 129 (filteredClassValues.Count == 0  !maxDensityClassValue.IsAlmost(filteredClassValues.Last()))) { 130 filteredThresholds.Add(thresholdList[i]); 131 filteredClassValues.Add(maxDensityClassValue); 132 } 133 } 134 thresholds = filteredThresholds.ToArray(); 135 classValues = filteredClassValues.ToArray(); 136 } 131 filteredThresholds.Insert(0, double.NegativeInfinity); 132 filteredClassValues.Insert(0, mostFrequentClass); 133 } 134 135 thresholds = filteredThresholds.ToArray(); 136 classValues = filteredClassValues.ToArray(); 137 137 } 138 138 … … 208 208 double g = Math.Sqrt(2 * s2 * s2 * Math.Log(s2 / s1)  2 * s1 * s1 * Math.Log(s2 / s1)  2 * m1 * m2 + m1 * m1 + m2 * m2); 209 209 double s = (s1 * s1  s2 * s2); 210 x1 = 210 x1 = (m2 * s1 * s1  m1 * s2 * s2 + s1 * s2 * g) / s; 211 211 x2 = (m1 * s2 * s2  m2 * s1 * s1 + s1 * s2 * g) / s; 212 212 }
Note: See TracChangeset
for help on using the changeset viewer.