What Is a Data Set? Definition and a Tabular Example

A data set, also written dataset, is a collection of related data treated as a unit. For spreadsheet work, that usually means a table of records with consistently defined columns—for example, monthly visits to library branches.

The term is broader than spreadsheets. W3C’s Data Catalog Vocabulary includes collections of numbers, text, images, sound and other data. It also distinguishes the dataset itself from the files used to distribute it: one dataset can be available in multiple formats. (W3C DCAT)

If you are preparing a spreadsheet for publication, the practical question is: What does each row represent, and what does each column mean?

A simple tabular dataset

Here is a fictional dataset for illustration—not real library statistics:

This dataset has four data rows and four columns. The header contains labels, not an observation.

• Row / record / observation: One branch in one calendar month.
• Column / field / variable: A property recorded for each observation, such as the month or visitor count.
• Cell / value: One entry at the intersection of a row and column, such as 1240.
• Identifier: branch_id identifies the branch. Together, branch_id and month identify an observation in this example.

These relationships match W3C’s tabular model: each row describes something, and cells in the same column describe the same property. (W3C tabular data model)

The distinction between 0 and the blank matters. In this fictional example, 0 means a confirmed count of zero recorded visits; the blank means the count was not reported. A reader should not have to guess that convention. See empty fields, zero and missing values.

Dataset, table and file are different concepts

A dataset is the collection you are describing. A table is a row-and-column structure used to organize it. A file is a container or representation, such as CSV or XLSX.

For example, a CSV export and an XLSX workbook can represent the same library-visit dataset if they contain the same records and values. But a workbook might also contain unrelated tables, notes and charts; the workbook is not automatically one coherent dataset. For format choices, see CSV vs XLSX.

Define the dataset before publishing it

Use a short description alongside the public table. For the sample above:

Monthly library visits — fictional example. Each row represents one branch in one calendar month. visits is a count of recorded visits, not unique visitors. Zero means no visits were recorded; a blank means the count was not reported.

For a real release, add the source, coverage period, publication or update date, units, collection method, license or reuse terms, and known limitations. W3C’s tabular model explicitly supports descriptions, source information and provenance as metadata associated with tables. (W3C)

Before sharing:

1. Check the row definition. Do not mix individual branches and system-wide totals as though they were equivalent observations.
2. Check the structure. Use clear headers, consistent values and documented missing-value markers. Verify the first observation and total row count.
3. Review the release copy. Remove sensitive information and confirm that you are authorized to publish the data.
4. Check the published result. Confirm that headers, dates, identifiers, zeros and blanks still mean what they meant in the source file.

As of October 24, 2026, TablePage accepts CSV, TSV, XLSX and XLS files and creates a public dataset page with a shareable link and filterable table. Upload only a reviewed, public-safe copy. (TablePage)