Product

Synthetic Australian population data

Privacy-safe synthetic individuals you can slice, model, simulate, or join to your own data. CSV / Excel / Parquet. 15,343 suburbs. 27.5 million synthetic persons.

For ML / AI teams

AUSynth is privacy-preserving synthetic Australian Census data, ideal for: training ML models without privacy risk, bias testing across demographic profiles, realistic test data for development / staging.

For large training runs, bundle the 5,000-credit pack — that covers 2.5M synthetic individuals at the standard rate.

For AI/ML teams →Free sample on Hugging Face →verosynthea-validator on GitHub →

How it works

Pick a geography (suburb / postcode / LGA / state / national), pick a dataset type (persons / families / dwellings), pick a format, and download. Each record is statistically representative of real ABS Census 2021, but is not a real Australian.

Open the query builder →See a free sample →

Pricing

1 credit per 500 records (fractional, so 100 records = 0.2 credits). Hierarchical-tier downloads (persons within families, etc.) cost 1.5× the base. See bundles →

What's in each record

47 Census variables, including age, sex, income, education, occupation, industry, household composition, dwelling type, and more. See the data dictionary for the full schema.

Worked examples

Three published reports built on the same data. Skim them to see what the variables look like when you actually ask a question with them.

Who uses this

EDUCATION

Realistic Australian data for student tutorials.Data

RESEARCH

Synthetic cohorts for simulation studies.Data

ML / DATA SCI

Train models on realistic data, zero privacy risk.Data

DEV / DEMO

Realistic test data for staging environments.Data
Customer demos without exposing real records.Data

HEALTH PLAN

Demographic mix for grant applications.Data

FINANCE

Income distributions for risk modelling.Data