experiment-results-planning

1# Experiment Results Planning
2 
3This skill designs the experiment/result layer before final metrics exist. It may generate mock planning data, but never presents mock data as real experimental evidence.
4 
5## Hard Gate
6 
7Before writing Results or Discussion, create:
8 
9- `plan/experiment-protocol.md`
10- `plan/review/method-experiment-traceability.md`
11- `tables/table-schema.md`
12- `figures/data-manifest.md`
13- real data files or clearly labeled `mock_*` files
14 
15## Experiment Protocol
16 
17The protocol must include:
18 
19- Dataset and split strategy.
20- Baselines and why each is fair.
21- Metrics and imbalance handling.
22- Main comparison.
23- Efficiency evaluation.
24- Ablation studies for each claimed module.
25- Generalization or robustness checks.
26- Explainability evaluation if XAI is a contribution.
27 
28Each contribution in Introduction must map to at least one experiment or limitation note.
29 
30## Recommended Experiment Gates
31 
32Use these gates in `plan/stage-gates.md` for result-heavy papers:
33 
341. Gate D0: Experiment Protocol Locked
35   - Required: datasets, split rules, Non-IID construction, seeds, baselines, metrics, hardware/software, log schema.
362. Gate D1: Method-Experiment Traceability
37   - Required: `plan/review/method-experiment-traceability.md`.
38   - Map each contribution to method modules, experiments, tables/figures, and allowed claims.
393. Gate D2: Table/Figure Data Contract
40   - Required: `tables/table-schema.md`, `figures/data-manifest.md`, and data files.
41…