Ranking

NotePlease be aware

We will monitor submissions for attempts to learn from or exploit the ranking process. We consider such practices unfair and inconsistent with the purposes and goals of the Data Challenge.

Leaderboard

ImportantImportant

Observable migrated to its new Notebook framework exactly on 1st Sep…which broke the leaderboard below and the plot of registered teams in the Stats page.
I am trying to fix it.
You can retrieve all submissions info via the API as explained below.


We will post urgent and important news news here, but we strongly recommend following the updates on the OSN Discord channel, especially #prc-data-competition

If you haven’t joined the OSN Discord server yet, you’ll need to do so first to access the channel.

All paginated submission rankings can be retrieved via the REST API at the full leaderboard JSON (use the value of nextCursor in the cursor= parameter in the URL query).

Other endpoints are also available; have a look here and use competitionId = bb3693e1-26bc-4a9e-8619-4fe78b4eab0c.

Submission Instructions

Teams will be ranked based on the best Root Mean Square Error (RMSE) achieved across their submissions.

The submitting.parquet file is provided as a template. To make a submission, fill in the TAXITIME_SEC_mvt column with your predicted taxi-out times for the corresponding MVT_ID_mvt values.

Each team must submit its predictions to its own bucket using the following file naming convention:

<team-name>_v<incremental integer>.parquet

For a submission to be valid, each row must match the corresponding MVT_ID_mvt before the ranking script can use the prediction to calculate the RMSE.

The ranking script will issue an error if:

  • an MVT_ID_mvt does not match;
  • any rows are missing; or
  • extra rows are included.

There is currently a limit of 3 submissions per day and a total max of 1GB per bucket.

List of Acronyms

RMSE: Root Mean Square Error