Purpose

Audit, clean, summarize, and prepare mathematical modeling contest data for downstream modeling.

This skill converts raw contest data and method data requirements into cleaned, traceable modeling data. It identifies data fields, checks data quality, handles missing values and anomalies with explicit rules, generates basic exploratory summaries, and produces artifacts that downstream code and paper sections can trust.

This skill does not select models, implement the main model, run final experiments, perform robustness analysis, or write paper sections.

When to use

Use this skill:

• After method-selector has produced a validated method plan.
• When raw data files are available under workspace/data/data_raw/.
• When data fields, units, missing values, abnormal values, or encodings need to be audited.
• Before model-code-analyzer.
• Before writing paper claims based on data.
• When the team needs a reproducible data cleaning record.