Sdílet prostřednictvím


MulticlassClassificationCatalog Třída

Definice

Třída používaná MLContext k vytváření instancí vícetřídních klasifikačních komponent, jako jsou trenéři.

public sealed class MulticlassClassificationCatalog : Microsoft.ML.TrainCatalogBase
type MulticlassClassificationCatalog = class
    inherit TrainCatalogBase
Public NotInheritable Class MulticlassClassificationCatalog
Inherits TrainCatalogBase
Dědičnost
MulticlassClassificationCatalog

Vlastnosti

Trainers

Seznam trenérů pro provádění vícetřídní klasifikace.

Metody

CrossValidate(IDataView, IEstimator<ITransformer>, Int32, String, String, Nullable<Int32>)

Spusťte křížové ověření překládáním numberOfFoldsdata, montáží estimatora dodržováním samplingKeyColumnName , pokud je k dispozici. Pak vyhodnoťte jednotlivé dílčí modely a labelColumnName vrátíte metriky.

Evaluate(IDataView, String, String, String, Int32)

Vyhodnotí data klasifikace s více třídami.

Metody rozšíření

PermutationFeatureImportance(MulticlassClassificationCatalog, ITransformer, IDataView, String, Boolean, Nullable<Int32>, Int32)

Důležitost funkce permutace (PFI) pro multiclassClassification

PermutationFeatureImportance<TModel>(MulticlassClassificationCatalog, ISingleFeaturePredictionTransformer<TModel>, IDataView, String, Boolean, Nullable<Int32>, Int32)

Důležitost funkce permutace (PFI) pro multiclassClassification

Platí pro