rel-f1 F1 Database

Database Description: The F1 database tracks all-time Formula 1 racing data and statistics since 1950. It provides detailed information for various stakeholders including drivers, constructors, engine manufacturers, and tyre manufacturers. Highlights include data on all circuits (e.g. geographical details), and full historical data from every season. This includes overall standings, race results, and more specific data like practice sessions, qualifying positions, sprints, and pit stops.

Database Statistics:

Database schema:

To load this relational database in RelBench, do:

from relbench.datasets import get_dataset
dataset = get_dataset("rel-f1")

References:

[1] F1DB.

Dataset License: CC-BY-4.0 license.

Entity Classification Tasks

driver-dnf

Task Description: For each driver predict the if they will DNF (did not finish) a race in the next 1 month.

Evaluation metric: AUROC

driver-top3

Task Description: For each driver predict if they will qualify in the top-3 for a race in the next 1 month.

Evaluation metric: AUROC

Entity Regression Tasks

driver-position

Task Description: Predict the average finishing position of each driver all races in the next 2 months.

Evaluation metric: MAE

results-position

Task Description: For each race result, predict the finishing position.

Evaluation metric: MAE

qualifying-position

Task Description: For each qualifying entry, predict the qualifying position.

Evaluation metric: MAE

constructor_results-points

Task Description: For each constructor's race result, predict the number of points scored in that race.

Evaluation metric: MAE

constructor_standings-position

Task Description: For each constructor standing, predict the constructor's position in the standings.

Evaluation metric: MAE

Link Prediction Tasks

driver-circuit-compete

Task Description: Predict in which circuits a driver will compete in the next 1 year.

Evaluation metricMAP