Write an analysis vignette

Use this shell for a real-data story, not an API reference. Keep the analytical pipe short; put acquisition, checks and plotting around it. See the collection contract for notebook heading equivalents.

Question

Ask one question that the available observations can address. State whether this is an executable notebook, a live-data recipe, or a dependency design.

Grammar / pipeline

Show the short pipe, with links to the canonical nouns and verbs. Do not duplicate their parameter tables.

Plain-language interpretation

Read the pipe left to right. Explain what each step means for this question.

Analysis

Supply complete setup, acquisition and analysis code. Declare optional dependencies and network requirements. Check units, dimensions, CRS, missing values and temporal coverage before combining data. Never invent observations.

Result

Plot the computed result, label units and axes, and explain the pattern. Distinguish a descriptive statistic from causal inference or a decision rule. If execution is blocked, say so instead of presenting an expected plot as output.

Data used

Record provider/product, variable, location, dates, resolution, processing, source revision or snapshot checksum, and limitations. Link to canonical source facts and citation guidance.

Reproduce

Give exact setup and run commands, expected outputs and verification steps. Declare what was actually executed, including whether live access was checked. For supported notebooks, run python scripts/run_vignettes.py and mkdocs build --strict.