Analytics Data Sets: Choose a Source and Build a Public Table

Choose analytics data sets by the question you want readers to explore—not by download size. For a first public table, a manageable task is to compare daily bike rentals by working-day status. The input is a documented CSV; the result is a table readers can filter by date and inspect alongside your summary.

An analytics dataset is a structured collection of observations used to compare groups, calculate measures or examine patterns. Before using one, establish what one row represents, what each measure means and whether you may republish it.

Three public datasets worth starting with

1. World Development Indicators: cross-country comparisons

The World Bank’s World Development Indicators covers development measures across themes including population, education, the economy and the environment. It offers bulk CSV and Excel downloads with metadata, as well as queries through DataBank.

Useful task: compare a selected indicator across economies over a defined period.

Suggested publication structure: one row per economy, year and indicator, with columns for economy code, indicator code, value and unit. This is a proposed release structure, not a promise about the layout of every download.

Select your countries, indicators and years before exporting; DataBank supports Excel, CSV, TXT and SDMX downloads.

Main check: use the same definition and observation year for a comparison. Do not silently rank one economy’s latest value against another’s older value, or mix country-group aggregates with individual economies.

The World Bank defaults to CC BY 4.0 for its own open datasets, but some datasets have different terms. Check the selected dataset’s metadata rather than treating everything hosted by the Bank as freely redistributable. Its licensing policy distinguishes open data, third-party data and microdata access conditions.

2. Our World in Data energy dataset: energy and electricity trends

The current energy dataset documentation describes one row per country or region and year, with columns for energy supply, electricity generation, energy mix and related measures. It also documents units and underlying sources for individual indicators.

Useful task: publish a country-year table comparing electricity generation from selected sources, keeping either absolute generation or percentage shares consistently labeled.

Main check: distinguish electricity from total energy supply, and countries from regional aggregates. A percentage of electricity generation is not a percentage of all energy use.

As of October 24, 2026, the repository directs users to a new release and marks the legacy dataset as no longer updated. The new release measures energy as total energy supply rather than using the previous substitution method; those energy measures are not directly comparable across releases. Build your table from one release rather than appending new figures to an old series without checking methodology and column changes.

For republication, credit Our World in Data and the underlying sources, and check the original providers’ licenses. The repository explicitly notes that third-party data retains its own license terms.

3. UCI Bike Sharing: a compact practice dataset

The UCI Bike Sharing dataset contains daily and hourly rental counts from the Capital Bikeshare system for 2011–2012, with weather and seasonal information. Its download includes day.csv, hour.csv and a readme. UCI lists it under CC BY 4.0, allowing sharing and adaptation with appropriate credit.

Useful task: compare daily rental counts between working and non-working days, separately by year.

Main check: these are historical, aggregated observations—not current service statistics or individual trip records. Weather fields such as temperature and humidity are normalized, so their stored values need explanation rather than being displayed as raw physical measurements.

This is the simplest starting point of the three if your goal is to practice preparing and publishing a CSV.

Worked release structure: daily bike rentals

Start with day.csv. Keep the original download unchanged and create a separate publication copy. The following is a proposed schema, using UCI’s documented fields:

Published column	Source or transformation	Meaning
date	dteday	Observation date
year	Decode yr: 0 → 2011; 1 → 2012	Calendar year
working_day	Decode workingday: 1 → Yes; 0 → No	Neither weekend nor holiday when Yes
casual_rentals	casual	Casual-user rental count
registered_rentals	registered	Registered-user rental count
total_rentals	cnt	Total rental count

UCI defines cnt as the total including casual and registered rentals. Use that relationship as a validation check: each row should satisfy casual_rentals + registered_rentals = total_rentals.

For a companion summary, group by year and working_day, then publish:

• Number of observed days.
• Total rentals.
• Mean daily rentals: total rentals divided by observed days.

Compare means rather than totals alone, because groups may contain different numbers of days. Describe any difference as an association; this grouping does not isolate the effects of weather, season or other changes.

Prepare the file and its public context

Before uploading:

1. Check structure. Use one header row, consistent column types and no subtotal rows mixed into observations. Check that dates are unique in this daily release.
2. Preserve missingness. Investigate unexpected blanks; do not replace them with zero merely to simplify a calculation. See empty fields versus zero.
3. Document provenance. Record the source link, creator, license, download date, coverage period, retained columns and transformations. Explain that non-working days include weekends and holidays.
4. Review public suitability. Publish only data whose reuse terms permit redistribution. Never upload sensitive information; removing names alone is not a sufficient privacy review.
5. Inspect the exported file. Reopen it and compare row counts, dates and totals with your reviewed spreadsheet. Keep the source documentation with your release records.

TablePage’s homepage currently lists CSV, TSV, XLSX and XLS uploads that generate public dataset pages with filterable tables. Upload only the reviewed publication copy. Then inspect the public page: check the first and last dates, numeric values and a representative filter before sharing its link. Check any displayed summaries against your own calculations.

Provide the definitions and provenance alongside the published table, wherever your publication supports that context. A clear caption and summary should tell readers what the rows represent, which period they cover and how to interpret the measures—not leave those answers inside the original download.