lenskit.data.MatrixRelationshipSet#

class lenskit.data.MatrixRelationshipSet(ds, name, schema, table)#

Bases: RelationshipSet

Two-entity relationships without duplicates, accessible in matrix form.

Note

Client code does not need to construct this class; obtain instances from a relationship set’s matrix() method.

Parameters:

ds (Dataset)
name (str)
schema (RelationshipSchema)
table (pa.Table)

__init__(ds, name, schema, table)#

Parameters:

ds (Dataset)
name (str)
schema (RelationshipSchema)
table (pa.Table)

Methods

`__init__`(ds, name, schema, table)
`arrow`(*[, attributes, ids])	Get these relationships and their attributes as a PyArrow table.
`col_stats`()
`coo_structure`()	Get the compressed sparse row structure of this relationship matrix.
`count`()
`csr_structure`()	Get the compressed sparse row structure of this relationship matrix.
`matrix`(*[, combine])	Convert this relationship set into a matrix, coalescing duplicate observations.
`pandas`(*[, attributes, ids])	Get these relationship and their attributes as a PyArrow table.
`row_items`([id, number])	Get a single row of this interaction matrix as an item list. Only valid when the column entity class is ``item''.
`row_stats`()
`row_table`([id, number])	Get a single row of this interaction matrix as a table.
`sample_negatives`(rows, *[, weighting, ...])	Sample negative columns (columns with no observation recorded) for an array of rows.
`scipy`([attribute, layout, legacy])	Get this relationship matrix as a SciPy sparse matrix.
`torch`([attribute, layout])	Get this relationship matrix as a PyTorch sparse tensor.

Attributes

`attribute_names`
`is_interaction`	Query whether these relationships represent interactions.
`n_cols`
`n_rows`
`row_vocabulary`
`row_type`
`col_vocabulary`
`col_type`
`rc_index`
`name`	The name of the relationship class for these relationships.
`schema`

matrix(*, combine=None)#

Convert this relationship set into a matrix, coalescing duplicate observations.

Parameters:: combine (Literal['count', 'sum', 'mean', 'first', 'last'] | dict[str, ~typing.Literal['count', 'sum', 'mean', 'first', 'last']] | None) – The method for combining attribute values for repeated relationships or interactions. Can either be a single strategy or a mapping from attribute names to combination strategies.
Return type:: MatrixRelationshipSet

csr_structure()#

Get the compressed sparse row structure of this relationship matrix.

Return type:: CSRStructure

coo_structure()#

Get the compressed sparse row structure of this relationship matrix.

Return type:: COOStructure

scipy(attribute=None, *, layout='csr', legacy=False)#

Get this relationship matrix as a SciPy sparse matrix.

Parameters:

attribute (str | None) – The attribute to return, or None to return an indicator-only sparse matrix (all observed values are 1).
layout (Literal['csr', 'coo']) – The matrix layout to return.
legacy (bool)

Returns:

The sparse matrix.

Return type:

sparray | spmatrix

torch(attribute=None, *, layout='csr')#

Get this relationship matrix as a PyTorch sparse tensor.

Parameters:

attribute (str | None) – The attribute to return, or None to return an indicator-only sparse matrix (all observed values are 1).
layout (Literal['csr', 'coo']) – The matrix layout to return.

Returns:

The sparse matrix.

Return type:

Tensor

sample_negatives(rows, *, weighting='uniform', verify=True, max_attempts=10, rng=None)#

Sample negative columns (columns with no observation recorded) for an array of rows. On a normal interaction matrix, this samples negative items for users.

Parameters:

rows (ndarray[int, dtype[int32]]) – The row numbers. Duplicates are allowed, and negative columns are sampled independently for each row. Must be a 1D array or tensor.
weighting (Literal['uniform', 'popularity']) – The weighting for sampled negatives; uniform samples them uniformly at random, while popularity samples them proportional to their popularity (number of occurrences).
verify (bool) – Whether to verify that the negative items are actually negative. Unverified sampling is much faster but can return false negatives.
max_attempts (int) – When verification is on, the maximum attempts before giving up and returning a possible false negative.
rng (Generator | None) – A random number generator to use.

Return type:

ndarray[Any, dtype[int32]]

row_table(id=None, *, number=None)#

Get a single row of this interaction matrix as a table.

Parameters:

id (int | str | bytes | integer[Any] | str_ | bytes_ | object_ | None)
number (int | None)

Return type:

Table | None

row_items(id=None, *, number=None)#

Get a single row of this interaction matrix as an item list. Only valid when the column entity class is ``item’’.

Parameters:

id (int | str | bytes | integer[Any] | str_ | bytes_ | object_ | None)
number (int | None)

Return type:

ItemList | None