---
name: jupyter-notebook-analysis-plan
description: "A plan for a notebook analysis that states the question first, checks the data, picks a method, plans the charts and ends with caveats. Use when a data question needs a notebook plan with checks, method, charts and caveats laid out before anyone writes analysis code. Do not use for cleaning a specific dataset or building a production dashboard. Related topics: data, research."
---

# Jupyter Notebook Analysis Plan

A plan for a notebook analysis that states the question first, checks the data, picks a method, plans the charts and ends with caveats.

## When to use

Use this skill when a data question needs a notebook plan with checks, method, charts and caveats laid out before anyone writes analysis code. Do not use it for cleaning a specific dataset or building a production dashboard.

## Workflow

1. Ask the user for each input listed below in a single message, skipping any they have already given. Where options are listed, offer them, but accept other answers too.
2. Restate the question as a measurable metric with a comparison, so the analysis has a finish line.
3. List the data checks to run first, each with the result that should halt the work.
4. Pick one method with a reason and one simpler baseline, sized to the data and the time budget.
5. Define each chart by the question it answers and the axes it uses, so no chart is decorative.
6. Write the conclusion template and the caveats section, including what finding would reverse the answer.

## Inputs

- **Question to answer** (longer free text)
- **Data sources** (longer free text)
- **Known data issues** (longer free text)
- **Audience** (choose one: Executives, Analysts, Engineers, Customers)
- **Tooling** (choose one: pandas, polars, SQL plus Python, R)
- **Time budget** (choose one: One hour, Half a day, One day, A week)

## Output

A six-section notebook outline with the measurable question, data checks with stop conditions, method and baseline, chart list, conclusion template and caveats.

## Prompt template

Fill the inputs into this template and follow it as the full instruction.

### Jupyter Notebook Analysis Plan

Plan a notebook analysis before any code is written.

- Question to answer: {{Question to answer}}
- Data sources: {{Data sources}}
- Known data issues: {{Known data issues}}
- Audience: {{Audience}}
- Tooling: {{Tooling}}
- Time budget: {{Time budget}}

Structure the notebook as numbered sections. Section 1 restates the question as something measurable, with the metric and the comparison. Section 2 lists data checks to run first: row counts, missing values, duplicates, date ranges, units and join keys, each with the result that would stop the analysis. Section 3 chooses a method and says why it fits the question and the data size, plus one simpler baseline to compare with. Section 4 names each chart, the exact question it answers and the axis choices. Section 5 holds the conclusion template: the answer in one sentence, the evidence and the confidence. Section 6 lists caveats and what would change the conclusion. Fit the scope to the time budget.

## Pitfalls

- Starting with modelling before the data checks, so a duplicated join key quietly inflates every number.
- Planning charts that show data without answering the stated question.
- Writing a conclusion that states certainty the sample size cannot support.