lenskit.algorithms.als.common

Classes

ALSBase(features, *[, epochs, reg, ...])	Base class for ALS models.
TrainContext(label, matrix, left, right, ...)	Context object for one half of an ALS training operation.
TrainingData(users, items, ui_rates, iu_rates)	Data for training the ALS model.

class lenskit.algorithms.als.common.TrainContext(label, matrix, left, right, reg, nrows, ncols, embed_size, regI)

Bases: NamedTuple

Context object for one half of an ALS training operation.

Parameters:

• label (str)

• matrix (Tensor)

• left (Tensor)

• right (Tensor)

• reg (float)

• nrows (int)

• ncols (int)

• embed_size (int)

• regI (Tensor)

label: str

Alias for field number 0

matrix: Tensor

Alias for field number 1

left: Tensor

Alias for field number 2

right: Tensor

Alias for field number 3

reg: float

Alias for field number 4

nrows: int

Alias for field number 5

ncols: int

Alias for field number 6

embed_size: int

Alias for field number 7

regI: Tensor

Alias for field number 8

class lenskit.algorithms.als.common.TrainingData(users, items, ui_rates, iu_rates)

Bases: NamedTuple

Data for training the ALS model.

Parameters:

• users (Vocabulary)

• items (Vocabulary)

• ui_rates (Tensor)

• iu_rates (Tensor)

users: Vocabulary

User ID mapping.

items: Vocabulary

Item ID mapping.

ui_rates: Tensor

User-item rating matrix.

iu_rates: Tensor

Item-user rating matrix.

to(device)

Move the training data to another device.

class lenskit.algorithms.als.common.ALSBase(features, *, epochs=10, reg=0.1, rng_spec=None, save_user_features=True)

Bases: MFPredictor[Tensor]

Base class for ALS models.

Parameters:

• features (int)

• epochs (int)

• reg (float | tuple[float, float])

• rng_spec (Optional[SeedLike])

• save_user_features (bool)

users_: Vocabulary | None

Users in the model (length=:math:m).

items_: Vocabulary

Items in the model (length=:math:n).

user_features_: Tensor | None

The \(m \times k\) user-feature matrix.

item_features_: Tensor

The \(m \times k\) user-feature matrix.

abstract property logger: Logger

Overridden in implementation to provide the logger.

fit(data, **kwargs)

Run ALS to train a model.

Parameters:

• ratings – the ratings data frame.

• data (Dataset)

Returns:

The algorithm (for chaining).

fit_iters(data, *, timer=None, **kwargs)

Run ALS to train a model, yielding after each iteration.

Parameters:

• ratings – the ratings data frame.

• data (Dataset)

• timer (Stopwatch | None)

Return type:

Iterator[Self]

abstract prepare_data(data)

Prepare data for training this model. This takes in the ratings, and is supposed to do two things:

• Normalize or transform the rating/interaction data, as needed, for training.

• Store any parameters learned from the normalization (e.g. means) in the appropriate member variables.

• Return the training data object to use for model training.

Parameters:

data (Dataset)

Return type:

TrainingData

initialize_params(data)

Initialize the model parameters at the beginning of training.

Parameters:

data (TrainingData)

abstract initial_params(nrows, ncols)

Compute initial parameter values of the specified shape.

Parameters:

• nrows (int)

• ncols (int)

Return type:

Tensor

abstract als_half_epoch(epoch, context)

Run one half of an ALS training epoch.

Parameters:

• epoch (int)

• context (TrainContext)

Return type:

float

predict_for_user(user, items, ratings=None)

Compute predictions for a user and items.

Parameters:

• user – the user ID

• items (array-like) – the items to predict

• ratings (pandas.Series) – the user’s ratings (indexed by item id); if provided, they may be used to override or augment the model’s notion of a user’s preferences.

Returns:

scores for the items, indexed by item id.

Return type:

pandas.Series

abstract new_user_embedding(user, ratings)

Generate an embedding for a user given their current ratings.

Parameters:

ratings (Series)

Return type:

tuple[Tensor, float | None]

finalize_scores(user, scores, u_offset)

Perform any final transformation of scores prior to returning them.

Parameters:

• scores (Series)

• u_offset (float | None)

Return type:

Series