Changeset 9363 for branches/OaaS/HeuristicLab.Algorithms.DataAnalysis/3.4/RandomForest/RandomForestClassification.cs
- Timestamp:
- 04/16/13 13:13:41 (11 years ago)
- Location:
- branches/OaaS
- Files:
-
- 4 edited
Legend:
- Unmodified
- Added
- Removed
-
branches/OaaS
- Property svn:ignore
-
old new 21 21 protoc.exe 22 22 _ReSharper.HeuristicLab 3.3 Tests 23 Google.ProtocolBuffers-2.4.1.473.dll 23 24 packages
-
- Property svn:mergeinfo changed
- Property svn:ignore
-
branches/OaaS/HeuristicLab.Algorithms.DataAnalysis
- Property svn:mergeinfo changed
-
branches/OaaS/HeuristicLab.Algorithms.DataAnalysis/3.4
- Property svn:ignore
-
old new 5 5 *.vs10x 6 6 Plugin.cs 7 *.user
-
- Property svn:ignore
-
branches/OaaS/HeuristicLab.Algorithms.DataAnalysis/3.4/RandomForest/RandomForestClassification.cs
r8139 r9363 26 26 using HeuristicLab.Core; 27 27 using HeuristicLab.Data; 28 using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;29 28 using HeuristicLab.Optimization; 29 using HeuristicLab.Parameters; 30 30 using HeuristicLab.Persistence.Default.CompositeSerializers.Storable; 31 31 using HeuristicLab.Problems.DataAnalysis; 32 using HeuristicLab.Problems.DataAnalysis.Symbolic;33 using HeuristicLab.Problems.DataAnalysis.Symbolic.Regression;34 using HeuristicLab.Parameters;35 32 36 33 namespace HeuristicLab.Algorithms.DataAnalysis { … … 45 42 private const string NumberOfTreesParameterName = "Number of trees"; 46 43 private const string RParameterName = "R"; 44 private const string MParameterName = "M"; 45 private const string SeedParameterName = "Seed"; 46 private const string SetSeedRandomlyParameterName = "SetSeedRandomly"; 47 47 48 #region parameter properties 48 public I ValueParameter<IntValue> NumberOfTreesParameter {49 get { return (I ValueParameter<IntValue>)Parameters[NumberOfTreesParameterName]; }49 public IFixedValueParameter<IntValue> NumberOfTreesParameter { 50 get { return (IFixedValueParameter<IntValue>)Parameters[NumberOfTreesParameterName]; } 50 51 } 51 public IValueParameter<DoubleValue> RParameter { 52 get { return (IValueParameter<DoubleValue>)Parameters[RParameterName]; } 52 public IFixedValueParameter<DoubleValue> RParameter { 53 get { return (IFixedValueParameter<DoubleValue>)Parameters[RParameterName]; } 54 } 55 public IFixedValueParameter<DoubleValue> MParameter { 56 get { return (IFixedValueParameter<DoubleValue>)Parameters[MParameterName]; } 57 } 58 public IFixedValueParameter<IntValue> SeedParameter { 59 get { return (IFixedValueParameter<IntValue>)Parameters[SeedParameterName]; } 60 } 61 public IFixedValueParameter<BoolValue> SetSeedRandomlyParameter { 62 get { return (IFixedValueParameter<BoolValue>)Parameters[SetSeedRandomlyParameterName]; } 53 63 } 54 64 #endregion … … 62 72 set { RParameter.Value.Value = value; } 63 73 } 74 public double M { 75 get { return MParameter.Value.Value; } 76 set { MParameter.Value.Value = value; } 77 } 78 public int Seed { 79 get { return SeedParameter.Value.Value; } 80 set { SeedParameter.Value.Value = value; } 81 } 82 public bool SetSeedRandomly { 83 get { return SetSeedRandomlyParameter.Value.Value; } 84 set { SetSeedRandomlyParameter.Value.Value = value; } 85 } 64 86 #endregion 87 65 88 [StorableConstructor] 66 89 private RandomForestClassification(bool deserializing) : base(deserializing) { } … … 68 91 : base(original, cloner) { 69 92 } 93 70 94 public RandomForestClassification() 71 95 : base() { 72 96 Parameters.Add(new FixedValueParameter<IntValue>(NumberOfTreesParameterName, "The number of trees in the forest. Should be between 50 and 100", new IntValue(50))); 73 97 Parameters.Add(new FixedValueParameter<DoubleValue>(RParameterName, "The ratio of the training set that will be used in the construction of individual trees (0<r<=1). Should be adjusted depending on the noise level in the dataset in the range from 0.66 (low noise) to 0.05 (high noise). This parameter should be adjusted to achieve good generalization error.", new DoubleValue(0.3))); 98 Parameters.Add(new FixedValueParameter<DoubleValue>(MParameterName, "The ratio of features that will be used in the construction of individual trees (0<m<=1)", new DoubleValue(0.5))); 99 Parameters.Add(new FixedValueParameter<IntValue>(SeedParameterName, "The random seed used to initialize the new pseudo random number generator.", new IntValue(0))); 100 Parameters.Add(new FixedValueParameter<BoolValue>(SetSeedRandomlyParameterName, "True if the random seed should be set to a random value, otherwise false.", new BoolValue(true))); 74 101 Problem = new ClassificationProblem(); 75 102 } 103 76 104 [StorableHook(HookType.AfterDeserialization)] 77 private void AfterDeserialization() { } 105 private void AfterDeserialization() { 106 if (!Parameters.ContainsKey(MParameterName)) 107 Parameters.Add(new FixedValueParameter<DoubleValue>(MParameterName, "The ratio of features that will be used in the construction of individual trees (0<m<=1)", new DoubleValue(0.5))); 108 if (!Parameters.ContainsKey(SeedParameterName)) 109 Parameters.Add(new FixedValueParameter<IntValue>(SeedParameterName, "The random seed used to initialize the new pseudo random number generator.", new IntValue(0))); 110 if (!Parameters.ContainsKey((SetSeedRandomlyParameterName))) 111 Parameters.Add(new FixedValueParameter<BoolValue>(SetSeedRandomlyParameterName, "True if the random seed should be set to a random value, otherwise false.", new BoolValue(true))); 112 } 78 113 79 114 public override IDeepCloneable Clone(Cloner cloner) { … … 84 119 protected override void Run() { 85 120 double rmsError, relClassificationError, outOfBagRmsError, outOfBagRelClassificationError; 86 var solution = CreateRandomForestClassificationSolution(Problem.ProblemData, NumberOfTrees, R, out rmsError, out relClassificationError, out outOfBagRmsError, out outOfBagRelClassificationError); 121 if (SetSeedRandomly) Seed = new System.Random().Next(); 122 123 var solution = CreateRandomForestClassificationSolution(Problem.ProblemData, NumberOfTrees, R, M, Seed, out rmsError, out relClassificationError, out outOfBagRmsError, out outOfBagRelClassificationError); 87 124 Results.Add(new Result(RandomForestClassificationModelResultName, "The random forest classification solution.", solution)); 88 125 Results.Add(new Result("Root mean square error", "The root of the mean of squared errors of the random forest regression solution on the training set.", new DoubleValue(rmsError))); … … 92 129 } 93 130 94 public static IClassificationSolution CreateRandomForestClassificationSolution(IClassificationProblemData problemData, int nTrees, double r, 131 public static IClassificationSolution CreateRandomForestClassificationSolution(IClassificationProblemData problemData, int nTrees, double r, double m, int seed, 95 132 out double rmsError, out double relClassificationError, out double outOfBagRmsError, out double outOfBagRelClassificationError) { 133 if (r <= 0 || r > 1) throw new ArgumentException("The R parameter in the random forest regression must be between 0 and 1."); 134 if (m <= 0 || m > 1) throw new ArgumentException("The M parameter in the random forest regression must be between 0 and 1."); 135 136 alglib.math.rndobject = new System.Random(seed); 137 96 138 Dataset dataset = problemData.Dataset; 97 139 string targetVariable = problemData.TargetVariable; … … 102 144 throw new NotSupportedException("Random forest classification does not support NaN or infinity values in the input dataset."); 103 145 146 int info = 0; 147 alglib.decisionforest dForest = new alglib.decisionforest(); 148 alglib.dfreport rep = new alglib.dfreport(); ; 149 int nRows = inputMatrix.GetLength(0); 150 int nColumns = inputMatrix.GetLength(1); 151 int sampleSize = Math.Max((int)Math.Round(r * nRows), 1); 152 int nFeatures = Math.Max((int)Math.Round(m * (nColumns - 1)), 1); 104 153 105 alglib.decisionforest dforest; 106 alglib.dfreport rep; 107 int nRows = inputMatrix.GetLength(0); 108 int nCols = inputMatrix.GetLength(1); 109 int info; 110 double[] classValues = dataset.GetDoubleValues(targetVariable).Distinct().OrderBy(x => x).ToArray(); 111 int nClasses = classValues.Count(); 154 155 double[] classValues = problemData.ClassValues.ToArray(); 156 int nClasses = problemData.Classes; 112 157 // map original class values to values [0..nClasses-1] 113 158 Dictionary<double, double> classIndices = new Dictionary<double, double>(); … … 116 161 } 117 162 for (int row = 0; row < nRows; row++) { 118 inputMatrix[row, nCol s - 1] = classIndices[inputMatrix[row, nCols - 1]];163 inputMatrix[row, nColumns - 1] = classIndices[inputMatrix[row, nColumns - 1]]; 119 164 } 120 // execute random forest algorithm 121 alglib.df buildrandomdecisionforest(inputMatrix, nRows, nCols - 1, nClasses, nTrees, r, out info, out dforest, out rep);165 // execute random forest algorithm 166 alglib.dforest.dfbuildinternal(inputMatrix, nRows, nColumns - 1, nClasses, nTrees, sampleSize, nFeatures, alglib.dforest.dfusestrongsplits + alglib.dforest.dfuseevs, ref info, dForest.innerobj, rep.innerobj); 122 167 if (info != 1) throw new ArgumentException("Error in calculation of random forest classification solution"); 123 168 … … 126 171 relClassificationError = rep.relclserror; 127 172 outOfBagRelClassificationError = rep.oobrelclserror; 128 return new RandomForestClassificationSolution((IClassificationProblemData)problemData.Clone(), new RandomForestModel(d forest, targetVariable, allowedInputVariables, classValues));173 return new RandomForestClassificationSolution((IClassificationProblemData)problemData.Clone(), new RandomForestModel(dForest, targetVariable, allowedInputVariables, classValues)); 129 174 } 130 175 #endregion
Note: See TracChangeset
for help on using the changeset viewer.