Package org.grouplens.lenskit.baseline

Baseline predictors.

See: Description

Class Summary

Class	Description
ConstantItemScorer	Rating scorer that predicts a constant rating for all items.
FallbackItemScorer	Item scorer that combines a primary scorer with a baseline.
GlobalMeanRatingItemScorer	Rating scorer that predicts the global mean rating for all items.
GlobalMeanRatingItemScorer.Builder	A default builder used to create GlobalMeanPredictors.
ItemMeanRatingItemScorer	Rating scorer that returns the item's mean rating for all predictions.
ItemMeanRatingItemScorer.Builder	A builder to create ItemMeanPredictors.
LeastSquaresItemScorer	Baseline scorer using least-squares estimates of preferences, trained by gradient descent.
LeastSquaresItemScorer.Builder	The builder for the least squares predictor.
UserMeanItemScorer	Rating scorer that returns the user's average rating for all predictions.

Enum Summary

Enum	Description
ScoreSource	Enum expressing where a score came from in recommender that uses a baseline fallback.

Annotation Types Summary

Annotation Type	Description
BaselineScorer	Annotation for the baseline scorer of a stacked item scorer, or an item scorer used as a baseline in another component.
ConstantItemScorer.Value	Parameter: the value used by the constant scorer.
MeanDamping	Damping parameter for means in baseline predictors.
PrimaryScorer	The primary scorer for a FallbackItemScorer.
UserMeanBaseline	Baseline scores for user mean ratings.

Package org.grouplens.lenskit.baseline Description

Baseline predictors.

Baseline predictors are like rating predictors, but they provide an unboxed SparseVector-based interface and are guaranteed to be able to predict for all users and items. They are used for things like normalizations and starting points for iterative methods.