The Database That Ate the Census - Release 2
After a substantial round of tweaking, I've managed to assemble a more complete and non-broken release of the Australian 2006 Census (in formats that don't suck), focusing on the SQLite version.
This second release resolves my previous problems with the broken SQLite file, and supplements the 45 raw tables (and 7600 raw columns) with a table metadata table, a column metadata table, and a hierachy table that will allow the generalisation from the raw Census Collection Districts (around 200-1000 people) into larger areas such as Local Government Areas, Postal Areas, States, and so on.
In addition to the fixed file formatting issues and the metadata tables, I've also added in a permalink for the main gzipped SQLite file. Shortly after the OSDC conference and my sister's wedding next week, I'll be throwing together and uploading a ORDB::AU::Census2006 module to automate the process of fetching and ORM-bootstrapping the census data.