4 Data Sets to Analyze—and Turn Into a Useful Public Table

Start with a dataset that can answer one question and produce one useful table. For example, use daily bike rentals to compare demand on working and non-working days, then publish a monthly summary with rental counts, daily averages and a source note.

These four datasets offer documented fields and spreadsheet-friendly files. Choose by the analysis you want to practice—not by the size of the download.

Dataset	Best starting task	File to use	Proposed public table
UCI Bike Sharing	Grouping dates and calculating averages	day.csv	Monthly rentals by working-day status
Palmer Penguins	Comparing groups and handling missing values	penguins.csv	Body measurements by species and recorded sex
World Bank population	Reshaping annual data and comparing change	Selected CSV export	Population by economy and year
Our World in Data CO₂	Comparing totals with per-person measures	CSV or XLSX	Annual emissions for selected countries

1. UCI Bike Sharing: practice time-based summaries

The UCI Bike Sharing dataset contains hourly and daily Capital Bikeshare rental counts for 2011–2012, alongside weather and calendar variables. Its download includes day.csv, hour.csv and a readme; UCI lists the dataset under CC BY 4.0.

Question to analyze: How did average daily rentals differ between working and non-working days across months?

Start with day.csv. Keep dteday, workingday, casual, registered and cnt. The source defines workingday=1 as neither a weekend nor a holiday, and cnt as the total including casual and registered users. Consequently, use either cnt or casual + registered for total rentals—not all three added together.

Watch for: Encoded categories and normalized measurements. In particular, the source’s temperature column is normalized; do not label its raw values as degrees Celsius. Treat this as a historical comparison, not a description of current ridership or proof that working-day status caused a difference in rentals.

2. Palmer Penguins: practice group comparisons

Palmer Penguins provides measurements for 344 penguins from three species and three islands in Antarctica’s Palmer Archipelago. The simplified file contains eight columns, including species, island, bill dimensions, flipper length, body mass, sex and year. Download the project’s simplified CSV for a spreadsheet project; no R installation is necessary to use that file.

Question to analyze: How does median body mass differ by species and recorded sex?

Publish group labels, the number of records used in each calculation, and median mass in grams. Keep the sample size next to each statistic so readers can judge how much data supports it.

Watch for: The CSV uses NA for missing values. For this comparison, use records with both body mass and recorded sex available, and disclose how many records you excluded. Do not replace missing measurements with zero. Report these as summaries of the sampled penguins, not estimates for every penguin in Antarctica.

The documentation lists a CC0 data license, provides citations, and asks people interested in publishing the data to contact Kristen Gorman about analysis and final products. Review that additional-use guidance before releasing your project.

3. World Bank population: practice annual comparisons

Use the World Bank’s Population, total indicator, code SP.POP.TOTL, for a focused comparison. The indicator page lists its contributing sources and a CC BY 4.0 license. In DataBank, select your economies, indicator and years; the World Bank’s download guidance confirms Excel and CSV export options.

Question to analyze: Which of five selected economies had the largest percentage population change over the same ten-year interval?

Prepare one row per economy and year, with columns such as economy_code, economy_name, year and population_people. If the export places years in separate columns, reshape those columns into rows.

Watch for: Units and geographic scope. As of October 24, 2026, the indicator webpage displays its country table in thousands; check the downloaded file’s metadata rather than assuming it uses the same display scaling. Exclude regional and income-group aggregates from an economy ranking. The World Bank also uses “country” interchangeably with “economy,” which can include territories, as its classification guidance explains.

For change, calculate (end population / start population − 1) × 100 only where both endpoint values are available and the starting value is positive. This is percentage change over the entire interval, not an annual growth rate.

4. Our World in Data CO₂: practice metric selection

The Our World in Data CO₂ repository offers CSV and XLSX files with one row per location and year, plus a codebook describing each indicator’s units and sources.

Question to analyze: Does the ordering of selected countries change when ranked by total emissions rather than emissions per person?

Keep country, iso_code, year, co2 and co2_per_capita. According to the codebook, co2 is annual CO₂ emissions excluding land-use change, measured in million tonnes; co2_per_capita is measured in tonnes per person and also excludes land-use change.

Watch for: Different measures answer different questions. Neither field represents all greenhouse gases. Use a common year with both measures available, and select country rows explicitly rather than mixing them with regional aggregates. Save a dated copy because the repository is updated regularly. Its reuse guidance requires checking underlying third-party licenses and citing both Our World in Data and the original sources.

Turn the analysis into a public table

For the bike-sharing project, use this proposed summary header:

month,workingday,days_observed,total_rentals,mean_daily_rentals

Build the summary as follows:

1. Preserve the original file. Save the download, readme and retrieval date separately from your working copy.
2. Check the row structure. In day.csv, verify that dates are unique. Derive month from dteday in YYYY-MM format, keeping January 2011 separate from January 2012. Group by that month and working-day status.
3. Calculate comparable summaries. Within each group, count the observed days, sum cnt, and divide that sum by the day count. Monthly totals alone can conceal differences in the number of working and non-working days.
4. Document the result. Explain that each row summarizes one calendar month and one working-day category. Include the period, category definitions, source, license, calculations and any exclusions.
5. Export a release copy. Use one header row, consistent dates and explicit units. Check missing-value handling with the spreadsheet exploration checklist.

Only upload data you are permitted to make public, and never include sensitive information. TablePage accepts CSV, TSV, XLSX and XLS files and creates public, shareable dataset pages with filterable tables. After upload, compare the published row count, dates and numeric values with your release copy before sharing the link. Keep the source and interpretation notes with the table wherever you share or embed it.