プロンプト

Move from problem definition and data quality to testable recommendations.

Choose methods and explain assumptions, intervals and limitations.

Choose charts that clearly communicate trends and comparisons.

SQL, section-by-section explanations, assumptions and validation queries.

A quality report, cleaning rules, sample code and validation assertions.

An analysis plan, chart suggestions, findings summary and limitations.

A metric dictionary, calculation examples, lineage and definition disputes.

A dashboard outline, chart rationale, filtering rules and data checks.

Retention definitions, calculation drafts, cohort tables and interpretation limits.

An analysis procedure, results table, assumption checks and decision rationale.

A forecasting plan, backtest design, model comparison and usage limits.

A theme codebook, segmented summaries, anonymous quotes and action list.

A search strategy, screening table, evidence matrix and review outline.

A paper summary, evidence chain, ranked comments and questions for authors.

A literature matrix, gap candidates, feasibility assessment and validation plan.

Normalized BibTeX, change notes, duplicate candidates and verification needs.

An experiment design, evaluation matrix, run log template and risk checklist.

A pipeline flow, data contracts, job dependencies and monitoring plan.

Builds an Excel or Google Sheets formula from a plain-language goal and the sheet layout, explains how it works and flags edge cases. Use when you know the result you want but not the formula.

Decides whether extreme values in an analysis are errors or real, sets a rule you can defend, and reports results with and without them using robust methods where they fit.

Plans anonymisation or pseudonymisation of a dataset before sharing, classifying identifiers, choosing techniques and assessing re-identification and residual risk. Use before data leaves your team.

Decides whether non-normal data actually matters for the planned analysis, using plots on the right quantity, and picks a fix (do nothing, transform, robust errors, bootstrap or nonparametric).

Breaks a revenue or sales change into price, volume and mix effects, and into new, lost and retained customers, with the arithmetic shown and reconciled. Use to explain why revenue moved.

Finds which behaviours and attributes predict churn in customer data, simple comparisons first and a model only if justified, with an action and a test per driver. Use at subscription businesses.