- Timestamp:
- 08/17/15 16:35:47 (9 years ago)
- Location:
- branches/crossvalidation-2434
- Files:
-
- 6 edited
- 1 copied
Legend:
- Unmodified
- Added
- Removed
-
branches/crossvalidation-2434
- Property svn:mergeinfo changed
/trunk/sources merged: 12835-12837,12839,12844-12846,12851,12855,12868
- Property svn:mergeinfo changed
-
branches/crossvalidation-2434/HeuristicLab.Algorithms.DataAnalysis
- Property svn:mergeinfo changed
/trunk/sources/HeuristicLab.Algorithms.DataAnalysis merged: 12868
- Property svn:mergeinfo changed
-
branches/crossvalidation-2434/HeuristicLab.Algorithms.DataAnalysis/3.4/GradientBoostedTrees/GradientBoostedTreesAlgorithm.cs
r12632 r12869 233 233 // produce solution 234 234 if (CreateSolution) { 235 var surrogateModel = new GradientBoostedTreesModelSurrogate(problemData, (uint)Seed, lossFunction.ToString(), 236 Iterations, MaxSize, R, M, Nu, state.GetModel()); 237 235 238 // for logistic regression we produce a classification solution 236 239 if (lossFunction is LogisticRegressionLoss) { 237 var model = new DiscriminantFunctionClassificationModel(s tate.GetModel(),240 var model = new DiscriminantFunctionClassificationModel(surrogateModel, 238 241 new AccuracyMaximizationThresholdCalculator()); 239 242 var classificationProblemData = new ClassificationProblemData(problemData.Dataset, … … 245 248 } else { 246 249 // otherwise we produce a regression solution 247 Results.Add(new Result("Solution", new RegressionSolution(s tate.GetModel(), problemData)));250 Results.Add(new Result("Solution", new RegressionSolution(surrogateModel, problemData))); 248 251 } 249 252 } -
branches/crossvalidation-2434/HeuristicLab.Algorithms.DataAnalysis/3.4/GradientBoostedTrees/GradientBoostedTreesModel.cs
r12660 r12869 34 34 // this is essentially a collection of weighted regression models 35 35 public sealed class GradientBoostedTreesModel : NamedItem, IRegressionModel { 36 [Storable] 36 // BackwardsCompatibility3.4 for allowing deserialization & serialization of old models 37 #region Backwards compatible code, remove with 3.5 38 private bool isCompatibilityLoaded = false; // only set to true if the model is deserialized from the old format, needed to make sure that information is serialized again if it was loaded from the old format 39 40 [Storable(Name = "models")] 41 private IList<IRegressionModel> __persistedModels { 42 set { 43 this.isCompatibilityLoaded = true; 44 this.models.Clear(); 45 foreach (var m in value) this.models.Add(m); 46 } 47 get { if (this.isCompatibilityLoaded) return models; else return null; } 48 } 49 [Storable(Name = "weights")] 50 private IList<double> __persistedWeights { 51 set { 52 this.isCompatibilityLoaded = true; 53 this.weights.Clear(); 54 foreach (var w in value) this.weights.Add(w); 55 } 56 get { if (this.isCompatibilityLoaded) return weights; else return null; } 57 } 58 #endregion 59 37 60 private readonly IList<IRegressionModel> models; 38 61 public IEnumerable<IRegressionModel> Models { get { return models; } } 39 62 40 [Storable]41 63 private readonly IList<double> weights; 42 64 public IEnumerable<double> Weights { get { return weights; } } 43 65 44 66 [StorableConstructor] 45 private GradientBoostedTreesModel(bool deserializing) : base(deserializing) { } 67 private GradientBoostedTreesModel(bool deserializing) 68 : base(deserializing) { 69 models = new List<IRegressionModel>(); 70 weights = new List<double>(); 71 } 46 72 private GradientBoostedTreesModel(GradientBoostedTreesModel original, Cloner cloner) 47 73 : base(original, cloner) { 48 74 this.weights = new List<double>(original.weights); 49 75 this.models = new List<IRegressionModel>(original.models.Select(m => cloner.Clone(m))); 76 this.isCompatibilityLoaded = original.isCompatibilityLoaded; 50 77 } 51 78 public GradientBoostedTreesModel(IEnumerable<IRegressionModel> models, IEnumerable<double> weights) … … 64 91 // allocate target array go over all models and add up weighted estimation for each row 65 92 if (!rows.Any()) return Enumerable.Empty<double>(); // return immediately if rows is empty. This prevents multiple iteration over lazy rows enumerable. 66 93 // (which essentially looks up indexes in a dictionary) 67 94 var res = new double[rows.Count()]; 68 95 for (int i = 0; i < models.Count; i++) { -
branches/crossvalidation-2434/HeuristicLab.Algorithms.DataAnalysis/3.4/GradientBoostedTrees/RegressionTreeModel.cs
r12700 r12869 82 82 [Storable] 83 83 // to prevent storing the references to data caches in nodes 84 // seemingly it is bad (performance-wise) to persist tuples (tuples are used as keys in a dictionary) TODO 84 85 private Tuple<string, double, int, int>[] SerializedTree { 85 86 get { return tree.Select(t => Tuple.Create(t.VarName, t.Val, t.LeftIdx, t.RightIdx)).ToArray(); } -
branches/crossvalidation-2434/HeuristicLab.Algorithms.DataAnalysis/3.4/HeuristicLab.Algorithms.DataAnalysis-3.4.csproj
r12700 r12869 195 195 <Compile Include="GaussianProcess\GaussianProcessRegressionSolution.cs" /> 196 196 <Compile Include="GaussianProcess\ICovarianceFunction.cs" /> 197 <Compile Include="GradientBoostedTrees\GradientBoostedTreesModelSurrogate.cs" /> 197 198 <Compile Include="GradientBoostedTrees\GradientBoostedTreesAlgorithm.cs" /> 198 199 <Compile Include="GradientBoostedTrees\GradientBoostedTreesAlgorithmStatic.cs" />
Note: See TracChangeset
for help on using the changeset viewer.