MyMediaLite  3.07
Class List
Here are the classes, structs, unions and interfaces with brief descriptions:
AsymmetricCorrelationMatrixClass for computing and storing correlations and similarities
AttributeDataClass that offers static methods to read (binary) attribute data into IBooleanMatrix objects
AUCArea under the ROC curve (AUC) of a list of ranked items
BiasedMatrixFactorizationMatrix factorization with explicit user and item bias, learning is performed by stochastic gradient descent
BidirectionalConditionalProbabilityClass for storing and computing 'bi-directional' conditional probabilities
BinaryCosineClass for storing cosine similarities
BinaryDataAsymmetricCorrelationMatrixClass with commoin routines for asymmetric correlations that are learned from binary data
BinaryDataSymmetricCorrelationMatrixClass with common routines for symmetric correlations that are learned from binary data
BiPolarSlopeOneBi-polar frequency-weighted Slope-One rating prediction
BPRLinearLinear model optimized for BPR
BPRMFMatrix factorization model for item prediction (ranking) optimized for BPR
BPRMF_ItemMappingBPR-MF with item mapping learned by regularized least-squares regression
BPRMF_ItemMapping_OptimalItem attribute to latent factor mapping, optimized for BPR loss
BPRMF_ItemMappingKNNBPR-MF with item mapping learned by kNN
BPRMF_ItemMappingSVRBPR-MF with item mapping learned by support-vector regression (SVR)
BPRMF_MappingBase class for BPR-MF plus attribute-to-factor mapping
BPRMF_MappingBPR-MF with attribute-to-factor mapping
BPRMF_UserMappingUser attribute to latent factor mapping for BPR-MF, optimized for RMSE on the latent factors
BPRMF_UserMapping_OptimalUser attribute to latent factor mapping for BPR-MF, optimized for BPR loss
BPRSLIMSparse Linear Methods (SLIM) for item prediction (ranking) optimized for BPR-Opt optimization criterion
CLiMFCollaborative Less-is-More Filtering Matrix Factorization
CoClusteringCo-clustering for rating prediction
CombinedList< T >Combines two List objects
CombinedRatingsCombine two IRatings objects
ConditionalProbabilityClass for storing and computing conditional probabilities
ConstantUses a constant rating value for prediction
ConstantsStatic class containing constants used by the MyMediaLite Input/Output routines
CooccurrenceClass for storing and computing the co-counts
DataReaderExtensionsExtension methods for IDataReader objects
DataSetAbstract dataset class that implements some common functions
EnsembleAbtract class for combining several prediction methods
EntityAverageAbstract class that uses an average (by entity) rating value for predictions
EntityMappingExtensionsI/O routines for classes implementing the IEntityMapping interface
EvaluationResultsClass for representing evaluation results
ExtensionsClass that contains static methods for rating prediction
ExtensionsHelper class with utility methods for handling recommenders
ExtensionsExtension methods for dataset statistics
ExtensionsClass that contains static methods for item prediction
ExtensionsExtension methods for correlation matrices
ExternalItemRecommenderUses externally computed predictions
ExternalRatingPredictorUses externally computed predictions
FactorWiseMatrixFactorizationMatrix factorization with factor-wise learning
FileSerializerStatic class for serializing objects to binary files
FoldInFold-in evaluation
FoldInRatingPredictorExtensionsExtension methods for IFoldInRatingPredictor
GlobalAverageUses the average rating value over all ratings for prediction
GridSearchGrid search for finding suitable hyperparameters
GSVDPlusPlusItem Attribute Aware SVD++: Matrix factorization that also takes into account _what_ users have rated and its attributes
HandlersClass containing handler functions, e.g. exception handlers
IBinaryDataCorrelationMatrixCorrelationMatrix that computes correlations over binary data
IBooleanMatrixInterface for boolean matrices
ICorrelationMatrixInterface representing correlation and similarity matrices
IDataSetInterface for different kinds of collaborative filtering data sets
IdentityMappingIdentity mapping for entity IDs: Every original ID is mapped to itself
IFoldInItemRecommenderItem recommender that allows folding in new users
IFoldInRatingPredictorRating predictor that allows folding in new users
IHyperParameterSearchInterface for classes that perform hyper-parameter search
IIncrementalItemRecommenderInterface for item recommenders
IIncrementalRatingPredictorInterface for rating predictors which support incremental training
IIncrementalRecommenderInterface for recommenders that support incremental model updates
IItemAttributeAwareRecommenderInterface for recommenders that take binary item attributes into account
IItemRelationAwareRecommenderInterface for recommenders that take a binary relation over items into account
IItemSimilarityProviderInterface for classes that provide item similarities
IIterativeModelInterface representing iteratively trained models
IMappingInterface to map external entity IDs to internal ones to ensure that there are no gaps in the numbering
IMatrix< T >Generic interface for matrix data types
IncrementalItemRecommenderBase class for item recommenders that support incremental updates
IncrementalRatingPredictorBase class for rating predictors that support incremental training
INeedsMappingsInterface for classes that need user and item ID mappings, e.g. for recommenders that read data from external sources and thus need to know which IDs are used externally
IPosOnlyFeedbackInterface for implicit, positive-only user feedback
IRatingCorrelationMatrixCorrelationMatrix that computes correlations over rating data
IRatingPredictorInterface for rating predictors
IRatingsInterface for rating datasets
IRecommenderGeneric interface for simple recommenders
ISplit< T >Generic dataset splitter interface
ItemAttributeKNNAttribute-aware weighted item-based kNN recommender
ItemAttributeKNNK-nearest neighbor (kNN) item-based collaborative filtering using the correlation of the item attibutes
ItemAttributeSVMContent-based filtering using one support-vector machine (SVM) per user
ItemAverageUses the average rating value of an item for prediction
ItemDataClass that contains static methods for reading in implicit feedback data for ItemRecommender
ItemDataRatingThresholdClass that contains static methods for reading in implicit feedback data for ItemRecommender, derived from rating data
ItemKNNWeighted item-based kNN
ItemKNNK-nearest neighbor (kNN) item-based collaborative filtering
ItemRecommendationEvaluationResultsItem recommendation evaluation results
ItemRecommenderAbstract item recommender class that loads the (positive-only implicit feedback) training data into memory and provides flexible access to it
ItemsEvaluation class for item recommendation
ItemsRoutines for reading in the item taxonomy of the KDD Cup 2011 data
ItemsCrossValidationCross-validation for item recommendation
ItemsOnlineOnline evaluation for rankings of items
ITimeAwareRatingPredictorInterface for time-aware rating predictors
ITimedDataSetInterface for data sets with time information
ITimedRatingsInterface for rating datasets with time information
ITransductiveItemRecommenderInterface for item recommenders that take into account some test data for training
ITransductiveRatingPredictorRating predictor that knows beforehand what it will have to rate
IUserAttributeAwareRecommenderInterface for recommenderss that take binary user attributes into account
IUserRelationAwareRecommenderInterface for recommenders that take a binary relation over users into account
IUserSimilarityProviderInterface for classes that provide user similarities
JaccardClass for storing and computing the Jaccard index (Tanimoto coefficient)
KDDCupItemsRepresents KDD Cup 2011 items like album, track, artist, or genre
KNNBase class for item recommenders that use some kind of k-nearest neighbors (kNN) model
KNNBase class for rating predictors that use some kind of kNN
LatentFeatureLogLinearModelLatent-feature log linear model
LeastSquareSLIMSparse Linear Methods (SLIM) for item prediction (ranking) optimized for the elastic net loss
ListProxy< T >Proxy class that allows access to selected elements of an underlying list data structure
LogisticLossUtility functions for the logistic loss
MAEUtility functions for the mean absolute error
MappingClass to map external entity IDs to internal ones to ensure that there are no gaps in the numbering
Matrix< T >Class for storing dense matrices
MatrixExtensionsUtilities to work with matrices
MatrixExtensionsUtilities to work with matrices
MatrixFactorizationSimple matrix factorization class, learning is performed by stochastic gradient descent (SGD)
MemoryMemory-related tools
MFAbstract class for matrix factorization based item predictors
ModelClass containing static routines for reading and writing recommender models
MostPopularMost-popular item recommender
MostPopularByAttributesRecommend most popular items by attribute
MovieLensRatingDataClass that offers static methods for reading in MovieLens 1M and 10M rating data
MultiCoreUtility routines for multi-core algorithms
MultiCoreBPRMFMatrix factorization for BPR on multiple cores
NaiveBayesAttribute-aware rating predictor using Naive Bayes
NDCGNormalized discounted cumulative gain (NDCG) of a list of ranked items
NelderMeadNealder-Mead algorithm for finding suitable hyperparameters
OverlapClass containing routines for computing overlaps
PearsonShrunk Pearson correlation for rating data
PosOnlyFeedback< T >Data structure for implicit, positive-only user feedback
PosOnlyFeedbackCrossValidationSplit< T >K-fold cross-validation split for item prediction from implicit feedback
PosOnlyFeedbackSimpleSplit< T >Simple split for item prediction from implicit feedback
PrecisionAndRecallPrecision and recall at different positions in the list
RandomRandom number generator singleton class
RandomUses a random rating value for prediction
RandomRandom item recommender for use as experimental baseline
RatingBasedRankingCrossValidationCross-validation for rating-based ranking
RatingCrossValidationSplitK-fold cross-validation split for rating prediction
RatingDataClass that offers methods for reading in rating data
RatingPredictionEvaluationResultsRating prediction evaluation results
RatingPredictorAbstract class for rating predictors that keep the rating data in memory for training (and possibly prediction)
RatingsEvaluation class for rating prediction
RatingsClass that offers static methods for reading in rating data from the KDD Cup 2011 files
RatingsData structure for storing ratings
RatingScaleClass containing information about the rating scale of a data set: valid rating values, minimum/maximum rating
RatingsChronologicalSplitChronological split for rating prediction
RatingsCrossValidationCross-validation for rating prediction
RatingsOnlineOnline evaluation for rating prediction
RatingsPerUserChronologicalSplitPer-user chronological split for rating prediction
RatingsProxyData structure that allows access to selected entries of a rating data structure
RatingsSimpleSplitSimple split for rating prediction
ReciprocalRankThe reciprocal rank of a list of ranked items
RecommenderAbstract recommender class implementing default behaviors
RecommenderParametersClass for key-value pair string processing
RelationDataClass that offers static methods to read (binary) relation over entities into IBooleanMatrix objects
RMSEUtility functions for the root mean square error (RMSE)
SequentialDiversificationSequential diversification
SigmoidCombinedAsymmetricFactorModelAsymmetric factor model which represents items in terms of the users that rated them, and users in terms of the items they rated
SigmoidItemAsymmetricFactorModelAsymmetric factor model
SigmoidSVDPlusPlusSVD++: Matrix factorization that also takes into account _what_ users have rated; variant that uses a sigmoid function
SigmoidUserAsymmetricFactorModelAsymmetric factor model which represents items in terms of the users that rated them
SkewSymmetricSparseMatrixSkew symmetric (anti-symmetric) sparse matrix; consumes less memory
SLIMAbstract class for SLIM based item predictors proposed by Ning and Karypis
SlopeOneFrequency-weighted Slope-One rating prediction
SocialMFSocial-network-aware matrix factorization
SoftMarginRankingMFMatrix factorization model for item prediction optimized for a soft margin (hinge) ranking loss, using stochastic gradient descent (as in BPR-MF)
SparseBooleanMatrixSparse representation of a boolean matrix, using HashSets
SparseMatrix< T >Class for storing sparse matrices
SparseMatrixExtensionsUtilities to work with matrices
StaticByteRatingsArray-based storage for rating data
StaticRatingDataClass that offers methods for reading in static rating data
StaticRatingsArray-based storage for rating data
SVDPlusPlusSVD++: Matrix factorization that also takes into account _what_ users have rated
SymmetricCorrelationMatrixClass for computing and storing correlations and similarities
SymmetricMatrix< T >Class for storing dense matrices
SymmetricSparseMatrix< T >Symmetric sparse matrix; consumes less memory
TimeAwareBaselineTime-aware bias model
TimeAwareBaselineWithFrequenciesTime-aware bias model with frequencies
TimeAwareRatingPredictorAbstract class for time-aware rating predictors
TimedRatingDataClass that offers methods for reading in rating data with time information
TimedRatingsData structure for storing ratings with time information
TimedRatingsProxyData structure that allows access to selected entries of a timed rating data structure
Track2ItemsClass that offers static methods for reading in test data from the KDD Cup 2011 files
TransductiveRatingPredictorExtensionsHelper methods for ITransductiveRatingPredictor
UserAttributeKNNWeighted kNN recommender based on user attributes
UserAttributeKNNK-nearest neighbor (kNN) user-based collaborative filtering using the correlation of the user attibutes
UserAverageUses the average rating value of a user for predictions
UserItemBaselineBaseline method for rating prediction
UserKNNK-nearest neighbor user-based collaborative filtering
UserKNNWeighted user-based kNN
UtilsClass containing utility functions
VectorExtensionsExtensions for vector-like data
VectorExtensionsExtensions for vector-like data
WeightedBPRMFWeigthed BPR-MF with frequency-adjusted sampling
WeightedEnsembleCombining several predictors with a weighted ensemble
WrapStatic methods to wrap around other code
WRMFWeighted matrix factorization method proposed by Hu et al. and Pan et al
ZeroConstant item recommender for use as experimental baseline. Always predicts a score of zero