Data Path · one question, one chart

Learn

Intro, terms, and tools

Data Path answers one question you care about in five phases: prepare a safe question, practice collecting rows, produce clean data and a chart, prove honesty with a second reader, present one insight with limits — private first.

Row is one record (one person, one day, one bus stop). Column is one type of info (hours slept, temperature, count). Bias is when the sample skews unfair — only your friends, only one season. Consent means people know what you collect and say OK; never store full names, emails, or student numbers.

Open Google Sheets. Bookmark Ontario open data for public CSVs. If you survey people, write how you will ask permission and anonymize before you collect row one.

You’ll know it when: question sentence + terms checked + Sheets open + open data bookmark.

If stuck: Question too big? Shrink until 15 rows fit one afternoon. Survey? Practice the consent line on a parent first.

Parent tip: Check the question is not exposing friends' private info — help anonymize early.

Learn

Pick a safe question

Write one question you can answer with a table — school sleep, club attendance, park litter counts, transit wait times, or a public dataset from Ontario’s open data portal. Kind and legal beats viral and creepy. If you survey people, get clear consent and never store full names, emails, or student numbers in the sheet.

Choose a source: your own log, a short survey, or a downloadable open-data CSV. Aim for 15+ clean rows. Add a note tab on how the data was collected and who it does (and doesn’t) represent.

You’ll know it when: sheet with 15 rows and a note on how data was collected.

If stuck: Shrink the question until collection fits one afternoon. Fifteen honest rows beat a hundred messy ones. Open data: filter to one city or one year to stay small.

Parent tip: Check the question isn’t exposing friends’ private info — help them anonymize early.

Learn

Clean, document, and chart

Rename columns so a stranger understands them without your narration. Fix blanks, typos, mixed units (minutes vs hours), and duplicate rows. Sort each important column once to spot impossible outliers — a −40°C room temperature usually means a sensor typo, not a discovery.

Write a one-line dictionary for each column on a second tab: what it means, units, and allowed values. Future-you (and a parent reviewer) will thank you. Note any rows you dropped and why.

Pick one chart type that fits the question — bar for categories, line for time, scatter for two numbers. Start axes at zero when a truncated axis would exaggerate a tiny change. Avoid 3D charts and rainbow spaghetti; clarity is the flex.

Ask someone else to read the chart cold with no preamble. Write a caption that states the actual finding (“Most responses were under 7 hours”) without hype. If they need a paragraph to understand it, simplify the chart or the question.

You’ll know it when: clean sheet + dictionary tab + one chart + caption stating the finding.

If stuck: Freeze the header row, then filter blanks. If a cell is mystery meat, leave a comment instead of guessing a number. If the chart needs a paragraph, drop a series or switch chart type. Export a PNG for later — polish can wait.

Parent tip: Ask “what’s the takeaway in one breath?” — then stop talking so they answer.

Learn

Prove the insight holds up

Ask someone else to read the chart cold with no preamble. Note what they think it says — if they need a paragraph, simplify.

Write a caption that states the actual finding ("Most responses were under 7 hours") without hype.

Name one limit of the data: sample size, season, who wasn't surveyed, or open-data quirks. Honest limits are strength.

You’ll know it when: second-reader note + caption + one limit sentence.

If stuck: Reader confused? Drop a series or switch chart type before present phase.

Parent tip: Ask "what's the takeaway in one breath?" — then stop talking so they answer.

Learn

Insight page with limits

Make a one-page insight: the question, the chart, a one-sentence finding, and one clear limit of the data (sample size, season, who wasn’t surveyed, open-data quirks). A Google Doc with a pasted chart counts — you are not required to publish a live website.

Share with family first. Only go wider if a parent says OK. End by saying what you would collect next time — curiosity without overclaiming is the skill.

You’ll know it when: insight page with chart, finding, and “what we don’t know yet.”

If stuck: A screenshot in a Doc is enough. If you feel pressure to “prove” more than the data shows, strengthen the limits section instead.

Parent tip: Celebrate honest limits — that’s real data science, not a weak finish.