research-design
Installation
SKILL.md
Research Design
Design experiments before running them. Bad design wastes months.
Workflow
Step 1 — Methodology Paradigm
Choose one and justify:
| Paradigm | When to use |
|---|---|
| Simulation | Can't control real environment; FL, scheduling, network protocols |
| Controlled experiment | Can manipulate variables; lab conditions |
| Real-world dataset | Benchmark comparison; image/NLP tasks |
| User study | Measuring human factors |
| Theoretical analysis | Proving bounds, convergence, complexity |
| Design Science Research (DSR) | Build + evaluate artifact in real context |