rel-event Event recommendation database

Database Description: The Event Recommendation database is obtained from user data on a mobile app called Hangtime. This app allows users to keep track of their friends' social plans. The database contains data on user actions, event metadata, and demographic information, as well as users' social relations, which captures how social relations can affect user behavior. Data is fully anonymized, with no personally identifiable information (such as names or aliases) available.

Database Statistics:

Database schema:

To load this relational database in RelBench, do:

from relbench.datasets import get_dataset
dataset = get_dataset("rel-event")

References:

[1] Event Recommendation Engine Challenge.

Dataset License: Open source upon authors agreement.

Node Classification Tasks

user-repeat

Task Description: Predict whether a user will attend an event(by responding yes or maybe) in the next 7 days if they have already attended an event in the last 14 days.

Evaluation metricAUROC

from relbench.datasets import get_dataset
dataset = get_dataset("rel-event")
task = dataset.get_task("user-repeat")
task.train_table, task.val_table, task.test_table # training/validation/testing tables

user-ignore

Task Description: Predict whether a user will ignore more than 2 event invitations in the next 7 days

Evaluation metricAUROC

from relbench.datasets import get_dataset
dataset = get_dataset("rel-event")
task = dataset.get_task("user-ignore")
task.train_table, task.val_table, task.test_table # training/validation/testing tables

Node Regression Tasks

user-attendance

Task Description: Predict how many events each user will respond yes or maybe in the next seven days.

Evaluation metricMAE

from relbench.datasets import get_dataset
dataset = get_dataset("rel-event")
task = dataset.get_task("user-attendance")
task.train_table, task.val_table, task.test_table # training/validation/testing tables