Datathon Deliverable Assessment Criteria

Published

May 24, 2026

The material submitted by each team will be assessed by a panel of experts based on the following criteria:

Scientific/Technical Accuracy

Definition: Measures correctness and rigor of the methods and outputs.

Key Points:

  • Appropriateness of modelling approach for the problem.
  • Correct handling of data (e.g., preprocessing, imputation, bias correction).
  • Logical and reproducible workflow from raw data to final output.
  • Validation against known or simulated benchmarks.

Scoring Tip: 0–5 scale; 5 = methodologically robust, fully validated, no obvious flaws.

Innovation and Creativity

Definition: Measures originality in problem-solving, methodology, or data integration.

Key Points:

  • Novel application of statistical, ML, or network techniques.
  • Creative integration of multiple data sources.
  • Unique visualization or decision-support approach.
  • Evidence of thinking beyond “textbook” solutions.

Scoring Tip: 0–5 scale; 5 = clearly innovative, methodologically or conceptually novel.

Practical Relevance / Policy Impact

Definition: Measures the degree to which outputs address WOAH priorities or real-world decision-making needs.

Key Points:

  • Alignment with animal health policy, surveillance, or trade risk objectives.
  • Interpretability for non-technical stakeholders.
  • Clarity of actionable insights or recommendations.
  • Responsiveness to intended problem scope.

Scoring Tip: 0–5 scale; 5 = deliverables are directly actionable and relevant.

Reproducibility and Documentation

Definition: Measures clarity and completeness of the workflow, code, and reporting.

Key Points:

  • Availability of scripts/notebooks and clear instructions.
  • Documentation of data sources, assumptions, and parameters.
  • Ability for independent evaluators to reproduce results.
  • Versioning or modular design supporting reuse.

Scoring Tip: 0–5 scale; 5 = fully reproducible and well-documented workflow.

Submission instructions

Submissions must be hosted as a private GitHub repository. Participants without a GitHub account should create one — it’s free.