Intersection

class rectools.metrics.intersection.Intersection(k: int, ref_k: Optional[int] = None)

Bases: MetricAtK

Metric to measure intersection in user-item pairs between recommendation lists.

The intersection@k equals the share of reco that is present in ref_reco.

This corresponds to the following algorithm:

1. filter reco by k

2. filter ref_reco by ref_k

3. calculate the proportion of items in reco that are also present in ref_reco

The second and third steps are equivalent to computing Recall@ref_k when:

• Interactions consists of reco without the Columns.Rank column.

• Recommendation table is ref_reco

Parameters

• k (int) – Number of items in top of recommendations list that will be used to calculate metric.

• ref_k (int, optional) – Number of items in top of reference recommendations list that will be used to calculate metric. If ref_k is None than ref_reco will be filtered with ref_k = k. Default: None.

Inherited-members

Methods

calc(reco, ref_reco)	Calculate metric value.
calc_per_user(reco, ref_reco)	Calculate metric values for all users.

Attributes

ref_k

calc(reco: DataFrame, ref_reco: DataFrame) → float

Calculate metric value.

Parameters

• reco (pd.DataFrame) – Recommendations table with columns Columns.User, Columns.Item, Columns.Rank.

• ref_reco (pd.DataFrame) – Reference recommendations table with columns Columns.User, Columns.Item, Columns.Rank.

Returns

Value of metric (average between users).

Return type

float

calc_per_user(reco: DataFrame, ref_reco: DataFrame) → Series

Calculate metric values for all users.

Parameters

• reco (pd.DataFrame) – Recommendations table with columns Columns.User, Columns.Item, Columns.Rank.

• ref_reco (pd.DataFrame) – Reference recommendations table with columns Columns.User, Columns.Item, Columns.Rank.

Returns

Values of metric (index - user id, values - metric value for every user).

Return type

pd.Series