Datathon

A datathon is a Single-Problem Competition variant with a data- science audience, characterized by a single dataset, a single objective metric, and leaderboard-driven comparison. Kaggle Days, the BMJ Datathon, and university datathons like UM Datathon are the paradigmatic examples. The mechanics are identical to the Single-Problem Competition archetype; the audience and tooling differ — data scientists rather than software engineers as the primary participants, Jupyter notebooks and ML pipelines as the primary delivery medium. The same named failure mode applies: the metric becomes a poor proxy when participants optimize for what gets measured rather than for what matters. See format-taxonomy.