Zerango editorial · October 16, 2026

A modern, reproducible data analysis workflow

Move from a question to cleaned data, SQL checks and an explanation of uncertainty.

Define the question

Choose one decision and identify the data that could inform it. Record where the data came from, what one row represents and which time range it covers. Do not start by producing every available chart.

Learn the data tools in sequence

Practice Python functions and collections before pandas. Then use DataFrames to inspect types, missing values, duplicated keys and inconsistent units. Preserve the input file and keep cleaning steps in a runnable script or notebook.

Use SQL to check assumptions

Practice filtering, grouping, joins and window functions with a small relational dataset. Compare row counts before and after a join. An unexpected increase can change an average or total even when the query runs without an error.

Explain the result

Choose a chart that answers the question, label the units and include the relevant comparison. Explain missing data and uncertainty. Keep correlation distinct from a claim about cause.

Use AI with executable checks

An assistant can suggest a query or a cleaning approach. Inspect the output, run it on a small example and test edge cases. The deliverable should include the original question, reproducible steps, a chart and a short decision memo.

Reviewed: October 16, 2026