Table: Training Data Click

The ml_reco__training_data_click table contains the offer click used during the Two Tower training.

Table description

name	data_type	description
event_date		The date when the event occurred, used for partitioning the data.
user_id		Unique identifier for a user.
item_id		Identifier for the item associated with the offer used internally by the data science team.
event_hour		The hour of the interaction event.
event_day		The day of the interaction event.
event_month		The month of the interaction event.
consult_offer		The number of distinct consultation occurences of an offer from home playlists on a given date.
booking_cnt		Total number of bookings, both individual and collective.
user_theoretical_remaining_credit		Total theoretical remaining credit for the user.
has_added_offer_to_favorites		Boolean. Indicates if the user has added any offer to their favorites.
user_queries		The search query associated with the event, if applicable.
qpi_subcategory_ids		QPI stands for 'Initial Practice Questionnaires'. We asked young users about their cultural practices before using the Pass, resulting in a list of subcategories used during the cold start to display offers based on these initial practices.
offer_subcategory_id		Identifier for the subcategory of the offer. Determined by the cultural partner via a list of pre-set options.
offer_category_id		Identifier for the category of the offer. Determined by the cultural partner via a list of pre-set options in a drop down menu.
item_image_embedding		Embedding generated with the offer image.
item_semantic_content_hybrid_embedding		Embedding generated with the image and the semantic content of an image.
item_names		Name of the offer as it appears in the application.
item_favourite_cnt		Total number of times this offer was favorited.