flurs.types.Event

class flurs.types.Event(user, item, value=1.0, context=array([], dtype=float64))[source]

An event object that represents a single user-item interaction.

Parameters

  • user (User) – A User instance.

  • item (Item) – An Item instance.

  • value (float, default=1.0) – A value representing the feedback. 1.0 in case of positive-only feedback, or 5-scale value for rating prediction, for example.

  • context (numpy array, default=empty) – Vector-represented contextual information associated with an interaction. An element can be day of the week, time, weather, etc.

__init__(user, item, value=1.0, context=array([], dtype=float64))[source]

Methods

__init__(user, item[, value, context])

encode([n_user, n_item, index, feature, ...])

Encode an event to an input vector for feature-based recommenders.
encode(n_user=None, n_item=None, index=True, feature=True, context=True, vertical=False)[source]

Encode an event to an input vector for feature-based recommenders.

Parameters

  • n_user (int) – Number of users currently registered to a recommender.

  • n_item (int) – Number of items currently registered to a recommender.

  • index (bool, default=True) – Include onehot-encoded user/item index to an input vector.

  • feature (bool, default=True) – Include features associated with user and item.

  • context (bool, default=True) – Include event-specific contextual information to a vector.

  • vertical (bool, default=False) – Return as a transposed n-by-1 vertical vector.

Returns

n-dimensional vector representing a user-item interaction. Size can be (n, 1) or (1, n), depending on vertical parameter.

Return type

array