Skip to content
Mock Address LabSouth Australia

Guided tour

The lab in three short walkthroughs

Each video follows one workflow from start to finish, with the step shown on screen, as captions and in the list beside it. A Playwright script recorded them from this site and checked every step on the way (the seeded sample, the interval and test statistic it quotes, the CSV it downloads, the suburb a lookup lands in), so a broken feature would fail the recording rather than appear in it.

Walkthrough 1 of 3 · /generate

Generate a test set

Two hundred mock addresses weighted by the config.py remoteness weights, with a fixed seed: generated in a Web Worker, mapped inside their suburb boundaries, and downloaded as CSV.

Setup: 200 addresses, Remoteness weights (the config.py defaults: 40 / 25 / 20 / 10 / 5%), seed 2025, coordinates on, no filters. The same settings give the same 200 addresses in any browser.Open the MP4

Steps (transcript)

  1. 1Generator: the 2025 recipe on the ABS 2021 suburb table, running in your browser
  2. 2Ask for 200 addresses with the fixed seed 2025: same seed, same list
  3. 3Weight by remoteness: the config.py weights the 2025 README promised but never applied
  4. 4Generate: 200 addresses in a Web Worker, each one stamped MOCK
  5. 5Map: every point is drawn inside its own suburb's ABS boundary
  6. 6Shade the suburbs by remoteness to see where the sample landed
  7. 7Choose CSV and download it: 200 rows with suburb, postcode, council and coordinates
Try it yourself

Walkthrough 2 of 3 · /generate

Did the sample hit the target?

The same 200 addresses checked against the mix they were drawn to hit: realised shares with 95% Wilson intervals and a goodness-of-fit test with its effect size, then the 2025 uniform generator against the weights its README promised, and the 200-seed study that shows the test is calibrated.

Setup: 200 addresses, seed 2025: first Remoteness weights, then Uniform (as built in 2025). Pearson chi-square test (every expected count is at least 5), Cohen's w for the effect size.Open the MP4

Steps (transcript)

  1. 1The same test set: 200 addresses, remoteness weights, seed 2025
  2. 2Target check: each area's realised share with a 95% Wilson interval; the dark tick is the target
  3. 3Major Cities: 84 of 200 (42%, 95% CI 35.4% to 48.9%) against a 40% target
  4. 4Pearson chi-square: χ²(4) = 2.77, p = 0.60, Cohen's w = 0.118 (small): on target
  5. 5Now the 2025 behaviour: uniform weighting, the same seed and the same 200
  6. 6Against the README's promised weights: χ²(4) = 74.68, p < 0.001, w = 0.611 (large): off target
  7. 7Show as a table: counts, Wilson intervals, targets and expected counts
  8. 8Over 200 seeds the test rejects the weighted design in 6 (3.0%, 95% CI 1.4% to 6.4%), near its 5% level
Try it yourself

Walkthrough 3 of 3 · /lookup

Look up a real address

Search a real place with Photon (OpenStreetMap) through this site's cached route, then find its suburb, postcode, council, remoteness area and SEIFA decile by point-in-polygon in the browser, from the city to the outback and by clicking the map.

Setup: Live Photon results for “Glenelg Jetty” and “Coober Pedy”; the suburb is found locally against the ABS 2021 boundaries, so a map click works without the geocoder.Open the MP4

Steps (transcript)

  1. 1Lookup: search a real South Australian place, or click the map
  2. 2Search “Glenelg Jetty”: Photon (OpenStreetMap) answers through this site's cached route
  3. 3Pick a result: point-in-polygon against the ABS 2021 boundaries, in your browser
  4. 4Glenelg: postcode 5045, Holdfast Bay council, Major Cities, IRSAD decile 8 in SA
  5. 5The jetty sits past the simplified coastline, so the nearest suburb is named, with the distance
  6. 6Coober Pedy, 751 km from the Adelaide GPO: Very Remote Australia, IRSAD decile 1 in SA
  7. 7Click anywhere on the map: the lookup works even when the geocoder is offline
Try it yourself

Screenshots

Every key feature at a glance

Captured by the same script, in light mode at 1440 × 900 (the landing page also in dark mode) and on a 390 px phone. Select one to enlarge it; the arrow keys step through the set.

Desktop · 1440 × 900

Mobile · 390 × 844

How these were made

pnpm showcase runs web/e2e/showcase.spec.ts from the repository on the system Chrome: it plays each journey at a human pace with an on-screen caption and cursor, asserts what it shows, and records it at 1280 × 800. ffmpeg then encodes the H.264 videos on this page and the GIFs in the README. The captions and the step lists here are the same text as the on-screen steps, and a unit test recomputes every number they quote.

No real API key is used anywhere. The AI screenshots use a placeholder key, every request to the provider is intercepted in the browser, and the reply is a labelled mock whose rationale starts with “Mocked response for illustration.”