cumcm-modeling-plan

name	cumcm-modeling-plan
description	Create rigorous modeling plans for 2026 CUMCM problems from problem statements, data descriptions, and constraints. Use when decomposing a CUMCM task, selecting candidate models, defining assumptions, metrics, algorithms, validation plans, sensitivity analysis, or contest-time modeling strategy.

CUMCM Modeling Plan

Use this skill before writing code or prose for a contest problem.

Workflow

1. Restate each subproblem as inputs, outputs, constraints, and evaluation criteria.
2. Propose 2-3 candidate model families with tradeoffs.
3. Select a primary path and a fallback path.
4. Specify:
   • variables and parameters
   • objective or loss function
   • required data transformations
   • solver or algorithm
   • validation and sensitivity tests
   • expected figures/tables
5. Read references/checklist.md when producing a full modeling plan.
6. Read references/problem-taxonomy.md before model selection, references/model-family-map.md for candidate methods, and validation/sensitivity references for review design.
7. Read references/examples.md when converting a problem statement or method comparison into an action plan.

Guardrails

• Separate assumptions from conclusions.
• Mark dependencies on unavailable data or ambiguous problem wording.
• Prefer robust, explainable methods over brittle novelty unless the problem rewards it.