publicdata.au publicdata.au on GitHub

About this site

publicdata.au is built and operated by National Digital, a software company in Burleigh Heads, Queensland. It is a private website. No government agency runs it, funds it or has endorsed it.

Why it exists

Australian governments publish a great deal of data under Creative Commons licences, and most of it arrives as a CSV or an Excel file on a portal. The file is replaced when the next release comes out, the URL often changes, and the history is gone. Anyone who wants JSON, Parquet or a database has to convert it themselves, and anyone who wants last year's release has to have kept a copy.

This site does that once, for everyone. Each release becomes a dated version at a URL that never changes. Each version carries its schema, its provenance and a diff against the release before. The publisher's own file sits beside it.

What it does to the data

Cells are typed. Headers are renamed to snake_case, with the original header kept in the schema. The encoding is made UTF-8. Blank cells become null. A cell the publisher suppressed, such as "<5", becomes null with a flag that says so. Rows are never added, removed, ranked, joined or summarised.

Licences

Each dataset is republished under the licence its publisher chose, and every file carries the attribution the licence asks for. A dataset whose licence does not allow derivative works is listed as blocked and never built. A dataset with no licence stated waits until the publisher confirms one.

The data licences are separate from the site's code, and stay that way. The code is open source under the GNU Affero General Public License, on GitHub, and the contributing page explains how to add a dataset or improve the site.

Corrections

If a value is wrong, it is almost always wrong in the publisher's file as well, because this site publishes what the publisher publishes. Send those to the publisher, whose contact is on each dataset page.

If the serialisation is wrong, for example a column typed badly or a row missing, open an issue on GitHub or tell National Digital. We fix it, publish a new build, and record the correction in the version's notes. A dated version keeps the same content once it is published. Its files change only to correct a fault in our conversion or in the publisher's attribution, to comply with the law, or when a publisher asks us to remove its dataset, and the change is recorded in that version's notes. The corrections policy sets out each step and keeps a log of every correction.

Citing the files

The licence on each dataset requires the publisher's attribution, and it is inside every file. We ask for one thing more: say that the file came from publicdata.au and link to the version you used. The link lets a reader fetch the same data, and it is how other people find this site. Every dataset page has the sentence ready to copy as text, HTML, Markdown and BibTeX.

Privacy

The site sets no cookies. Page views are counted by Cloudflare Web Analytics, which is served from this site's own provider, stores nothing in the browser and does not follow anyone across sites. There is no other tracking. Votes in the backlog are counted once per browser per day using a salted hash that changes every day. Nobody is asked who they are. The privacy page sets out everything the site records.

Security

The disclosure policy is at /.well-known/security.txt.