Class TrainTestSplittingRecommenderMapping<TInstanceSource, TInstance, TUser, TItem, TFeatureSource, TFeatureValues>

Represents the data splitting mapping that can be wrapped around any other recommender mapping to split the underlying data into the training and test parts.

Inheritance

Object

TrainTestSplittingRecommenderMapping<TInstanceSource, TInstance, TUser, TItem, TFeatureSource, TFeatureValues>

TrainTestSplittingRatingRecommenderMapping<TInstanceSource, TInstance, TUser, TItem, TRating, TFeatureSource, TFeatureValues>

Implements

IRecommenderMapping<SplitInstanceSource<TInstanceSource>, TInstance, TUser, TItem, TFeatureSource, TFeatureValues>

Inherited Members

Object.Equals(Object)

Object.Equals(Object, Object)

Object.GetHashCode()

Object.GetType()

Object.MemberwiseClone()

Object.ReferenceEquals(Object, Object)

Object.ToString()

Namespace: Microsoft.ML.Probabilistic.Learners.Mappings

Assembly: Microsoft.ML.Probabilistic.Learners.Recommender.dll

Syntax

public class TrainTestSplittingRecommenderMapping<TInstanceSource, TInstance, TUser, TItem, TFeatureSource, TFeatureValues> : IRecommenderMapping<SplitInstanceSource<TInstanceSource>, TInstance, TUser, TItem, TFeatureSource, TFeatureValues>

Type Parameters

Name	Description
TInstanceSource	The type of a source of instances.
TInstance	The type of an instance.
TUser	The type of a user.
TItem	The type of an item.
TFeatureSource	The type of a feature source.
TFeatureValues	The type of the feature values.

Remarks

The requested fraction of items is removed from the consideration (thus, ignored) with all the associated observations.
The requested fraction of items is put into the test set (so called cold items) with all the associated observations.
The requested fraction of the users remaining after (1-2) is removed from consideration (thus, ignored) with all the associated observations.
The requested fraction of the users remaining after (1-2) is put into the test set (so called cold users) with all the associated observations.
The requested fraction of the users remaining after (1-2) is put into the training set with all the associated observations.
For each user remaining after (1-5) the requested fraction of observations is put into the training set, while the rest is put into the test set. At least one observation is always put into the training set for each user.
If requested, instances associated with the occasionally produced cold items are removed from the test set. An item is said to be "occasionally cold" if it is covered only by the test set but wasn't explicitly chosen as cold. Such items can be produced by the steps (4) and (6). This option is supposed to be used when the requested number of cold users and items is set to zero, in order to ensure that all the entities in the test set are covered by the training set.

Constructors

TrainTestSplittingRecommenderMapping(IRecommenderMapping<TInstanceSource, TInstance, TUser, TItem, TFeatureSource, TFeatureValues>, Double, Double, Double, Double, Double, Double, Boolean)

Initializes a new instance of the TrainTestSplittingRecommenderMapping<TInstanceSource, TInstance, TUser, TItem, TFeatureSource, TFeatureValues> class.

Declaration

public TrainTestSplittingRecommenderMapping(IRecommenderMapping<TInstanceSource, TInstance, TUser, TItem, TFeatureSource, TFeatureValues> mapping, double trainingOnlyUserFraction, double testUserRatingTrainingFraction, double coldUserFraction = 0, double coldItemFraction = 0, double ignoredUserFraction = 0, double ignoredItemFraction = 0, bool removeOccasionalColdItems = false)

Parameters

Type	Name	Description
IRecommenderMapping<TInstanceSource, TInstance, TUser, TItem, TFeatureSource, TFeatureValues>	mapping	The wrapped recommender mapping.
Double	trainingOnlyUserFraction	The fraction of users included in the training set only.
Double	testUserRatingTrainingFraction	The fraction of ratings in the training set for each user who is presented in both sets.
Double	coldUserFraction	The fraction of users included in the test set only.
Double	coldItemFraction	The fraction of items included in the test set only.
Double	ignoredUserFraction	The fraction of users not included in either the training or the test set.
Double	ignoredItemFraction	The fraction of items not included in either the training or the test set.
Boolean	removeOccasionalColdItems	Whether the occasionally produced cold items should be removed from the test set.

Methods

GetInstances(SplitInstanceSource<TInstanceSource>)

Retrieves a list of instances from a given instance source. Dependent on IsTrainingSet, the instances of the training or test set are returned.

Declaration

public IEnumerable<TInstance> GetInstances(SplitInstanceSource<TInstanceSource> instanceSource)

Parameters

Type	Name	Description
SplitInstanceSource<TInstanceSource>	instanceSource	The source to retrieve instances from.

Returns

Type	Description
IEnumerable<TInstance>	The list of retrieved instances.

GetItem(SplitInstanceSource<TInstanceSource>, TInstance)

Extracts an item from a given instance by delegating the call to the wrapped mapping.

Declaration

public TItem GetItem(SplitInstanceSource<TInstanceSource> instanceSource, TInstance instance)

Parameters

Type	Name	Description
SplitInstanceSource<TInstanceSource>	instanceSource	The source of instances providing the `instance`.
TInstance	instance	The instance to extract the item from.

Returns

Type	Description
TItem	The extracted item.

GetItemFeatures(TFeatureSource, TItem)

Provides the features for a given item by delegating the call to the wrapped mapping.

Declaration

public TFeatureValues GetItemFeatures(TFeatureSource featureSource, TItem item)

Parameters

Type	Name	Description
TFeatureSource	featureSource	The source of the features.
TItem	item	The item to provide the features for.

Returns

Type	Description
TFeatureValues	The features for `item`.

GetUser(SplitInstanceSource<TInstanceSource>, TInstance)

Extracts a user from a given instance by delegating the call to the wrapped mapping.

Declaration

public TUser GetUser(SplitInstanceSource<TInstanceSource> instanceSource, TInstance instance)

Parameters

Type	Name	Description
SplitInstanceSource<TInstanceSource>	instanceSource	The source of instances providing the `instance`.
TInstance	instance	The instance to extract the user from.

Returns

Type	Description
TUser	The extracted user.

GetUserFeatures(TFeatureSource, TUser)

Provides the features for a given user by delegating the call to the wrapped mapping.

Declaration

public TFeatureValues GetUserFeatures(TFeatureSource featureSource, TUser user)

Parameters

Type	Name	Description
TFeatureSource	featureSource	The source of the features.
TUser	user	The user to provide the features for.

Returns

Type	Description
TFeatureValues	The features for `user`.

Implements

IRecommenderMapping<TInstanceSource, TInstance, TUser, TItem, TFeatureSource, TFeatureValues>

Extension Methods

RecommenderMappingExtensions.SplitToTrainTest<TInstanceSource, TInstance, TUser, TItem, TFeatureSource, TFeatureValues>(IRecommenderMapping<TInstanceSource, TInstance, TUser, TItem, TFeatureSource, TFeatureValues>, Double, Double, Double, Double, Double, Double, Boolean)

RecommenderMappingExtensions.WithGeneratedNegativeData<TInstanceSource, TInstance, TUser, TItem, TFeatureSource, TFeatureValues>(IRecommenderMapping<TInstanceSource, TInstance, TUser, TItem, TFeatureSource, TFeatureValues>, Double)