rel-avito Avito advertisement database
Database Description: Avito is a leading online advertisement platform, providing a marketplace for users to buy and sell a wide variety of products and services, including real estate, vehicles, jobs, and goods. The Avito Context Ad Clicks dataset on Kaggle is part of a competition aimed at predicting whether an ad will be clicked based on contextual information. This dataset includes user searches, ad attributes, and other related data to help build predictive models.
Database Statistics:
| Domain | E-commerce |
| Num of Tables | 8 |
| Num of Rows | 20,679,117 |
| Num of Columns | 43 |
| Starting time | 2015-04-25 |
| Validation timestamp | 2015-05-09 |
| Testing timestamp | 2015-05-14 |
| Time window size | 4 days |
Database schema:
To load this relational database in RelBench, do:
from relbench.datasets import get_dataset
dataset = get_dataset("rel-avito")
References:
[1] Avito Context Ad Clicks Challenge.
Dataset License: Kaggle license .
Node Classification Tasks
user-visits
Task Description: Predict whether each customer will visit more than one Ad in the next 4 days.
Evaluation metricAUROC
from relbench.datasets import get_dataset
dataset = get_dataset("rel-avito")
task = dataset.get_task("user-visits")
task.train_table, task.val_table, task.test_table # training/validation/testing tables
user-clicks
Task Description: Predict whether each customer will click on more than one Ads in the next 4 days.
Evaluation metricAUROC
from relbench.datasets import get_dataset
dataset = get_dataset("rel-avito")
task = dataset.get_task("user-clicks")
task.train_table, task.val_table, task.test_table # training/validation/testing tables
Node Regression Tasks
ad-ctr
Task Description: Assuming the Ad will be clicked in the next 4 days, predict the Click-Through-Rate (CTR) for each Ad.
Evaluation metricMAE
from relbench.datasets import get_dataset
dataset = get_dataset("rel-avito")
task = dataset.get_task("ad-ctr")
task.train_table, task.val_table, task.test_table # training/validation/testing tables
Link Prediction Tasks
user-ad-visit
Task Description: Predict the list of ads a user will visit in the next 4 days.
Evaluation metricMAP
from relbench.datasets import get_dataset
dataset = get_dataset("rel-avito")
task = dataset.get_task("user-ad-visit")
task.train_table, task.val_table, task.test_table # training/validation/testing tables