Namespace Microsoft.ML.Probabilistic.Learners

Classes

BayesPointMachineClassifier

The Bayes point machine classifier factory.

BayesPointMachineClassifierCapabilities

Defines the capabilities of the Bayes point machine classifier.

BayesPointMachineClassifierException

The exception that is thrown when the multi-class Bayes point machine classifier encounters an issue.

BayesPointMachineClassifierIterationChangedEventArgs

Provides information about the training progress of the Bayes point machine classifiers.

BayesPointMachineClassifierPredictionSettings<TLabel>

Abstract prediction settings of a Bayes point machine classifier.

BayesPointMachineClassifierSettings<TLabel, TTrainingSettings, TPredictionSettings>

Abstract settings of the Bayes point machine classifier.

BayesPointMachineClassifierTrainingSettings

Settings for the Bayes point machine classifier which affect training.

BinaryBayesPointMachineClassifierPredictionSettings<TLabel>

Settings for the binary Bayes point machine classifier which affect prediction.

BinaryBayesPointMachineClassifierSettings<TLabel>

Settings of the binary Bayes point machine classifier.

ClassifierEvaluator<TInstanceSource, TInstance, TLabelSource, TLabel>

Evaluates the predictions of a classifier.

ConfusionMatrix<TLabel>

Implements a confusion matrix.

DummyFeatureSource

Indicates that no explicit feature source is needed because features are implicitly stored somewhere else.

EntityPosteriorDistribution

Contains the learned parameters for an entity (user or item).

FeaturePosteriorDistribution

Represents the posterior distribution over feature weights.

ItemPosteriorDistribution

Contains the learned parameters for an item.

MatchboxRecommender

Matchbox recommender factory.

MatchboxRecommenderAdvancedTrainingSettings

Advanced settings of the Matchbox recommender which affect training. Cannot be set after training.

MatchboxRecommenderCapabilities

Defines the capabilities of the Matchbox recommender.

MatchboxRecommenderException

The exception that is thrown in the case of some issues encountered by the recommendation engine.

MatchboxRecommenderPredictionSettings

Settings of the Matchbox recommender which affect prediction.

MatchboxRecommenderSettings

Settings of the Matchbox recommender (settable by the developer).

MatchboxRecommenderTrainingSettings

Settings of the Matchbox recommender which affect training. Cannot be set after training.

Metrics

A diverse set of metrics to evaluate various kinds of predictors.

MulticlassBayesPointMachineClassifierPredictionSettings<TLabel>

Settings for the multi-class Bayes point machine classifier which affect prediction.

MulticlassBayesPointMachineClassifierSettings<TLabel>

Settings of the multi-class Bayes point machine classifier.

NoFeatureSource

Indicates that no feature source is needed because the model does not use features.

PointEstimator

Implements point estimators.

PosteriorDistributions<TUser, TItem>

The posterior distribution over the parameters of the Matchbox model.

RandomStarRatingRecommender<TInstanceSource, TInstance, TUser, TItem, TDataRating, TFeatureSource, TFeatureValues>

Represents a star rating recommender system which generates predictions purely by random guessing.

RandomStarRatingRecommenderCapabilities

Defines the capabilities of the Matchbox recommender.

RatingInstance<TUser, TItem, TRating>

Represents a user-item-rating triple.

RatingMatrix

Represents a matrix of values for predicted ratings versus true ratings. Usages include a confusion matrix and a loss matrix.

RecommenderEvaluator<TInstanceSource, TUser, TItem, TGroundTruthRating, TPredictedRating, TPredictedRatingDist>

Evaluates a recommender system.

RoundingStarRatingInfo

An implementation of IStarRatingInfo<TRating> which converts floating-point ratings to star ratings by rounding.

SerializationVersionAttribute

Sets the serialization version of the learner.

SettingsGuard

Guards settings from being changed.

StarRatingInfo

Provides a mapping for the case in which ratings are already star ratings.

StarRatingRecommenderEvaluator<TInstanceSource, TUser, TItem, TGroundTruthRating>

Evaluates a recommender system which predicts star ratings.

UserPosteriorDistribution

Contains the learned parameters for a user.

Utilities

Implements various utilities for all learners.

Structs

CalibrationPair

Struct which holds empirical and predicted probabilities for use in calibration.

FalseAndTruePositiveRate

Struct which holds both the FPR and TPR

PrecisionRecall

Struct which holds the precision and recall

Interfaces

IBayesPointMachineClassifier<TInstanceSource, TInstance, TLabelSource, TLabel, TLabelDistribution, TTrainingSettings, TPredictionSettings>

Interface to a Bayes point machine classifier.

IBayesPointMachineClassifierPredictionSettings<TLabel>

Interface to prediction settings of a Bayes point machine classifier.

IBayesPointMachineClassifierSettings<TLabel, TTrainingSettings, TPredictionSettings>

Interface to settings of a Bayes point machine classifier.

ICapabilities

Interface to learner capabilities.

ICustomSerializable

Interface to any object that needs to control its own serialization.

ILearner

Interface to a learner (something that can do machine learning).

IMatchboxRecommender<TInstanceSource, TUser, TItem, TRatingDistribution, TFeatureSource>

Interface to a Matchbox recommender system.

IPredictor<TInstanceSource, TInstance, TLabelSource, TResult, TResultDist>

Interface to a learner that acts on some data to predict a label.

IPredictorCapabilities

Interface to predictor capabilities.

IPredictorIncrementalTraining<TInstanceSource, TLabelSource>

Interface to a predictor which can be trained incrementally.

IRecommender<TInstanceSource, TUser, TItem, TRating, TRatingDistribution, TFeatureSource>

Interface to a recommendation algorithm.

IRecommenderCapabilities

Interface to a recommender capabilities.

ISettings

Interface to the settings of an implementation of ILearner. These should be set once to configure the learner before calling any query methods on it.

IStarRatingInfo<TRating>

Interface to provide a mapping from ratings of arbitrary type TRating to star ratings.

Enums

LossFunction

The loss function which determines how a prediction in the form of a distribution is converted into a point prediction.

RecommenderMetricAggregationMethod

Specifies how metrics are aggregated over the whole dataset.