Transport Inequality Engine

A Rust pipeline ingests the complete PTV GTFS feed and scores every stop in the network, train, tram, and bus alike, on a single 0 to 100 scale. Nothing is sampled or estimated from a subset of routes.

Averaging every stop score in a suburb sounds reasonable and produces misleading numbers: a handful of well-served stops near a station can flatten a suburb that is otherwise poorly served. TIE weights aggregation by where stops actually sit relative to where people live, so a suburb score reflects lived accessibility rather than stop density.

Suburb boundaries include parks, industrial land, and reserves that no one lives on. TIE applies dasymetric modelling to redistribute population within a suburb according to residential land use, then weights transit scores toward where residents concentrate.

Beyond area scores, TIE scores individual routes. That makes it possible to diagnose why a suburb underperforms: a specific bus route running infrequently, a tram line with poor weekend coverage, rather than a single opaque number.

Most transit accessibility models assume every trip radiates toward the CBD. TIE also measures orbital accessibility, suburb-to-suburb connections that skip the city centre, which is how most short trips actually work.

Address lookups happen entirely in the browser via OpenStreetMap Nominatim. No address is ever transmitted to or stored on a server. The backend only sees an aggregate count of lookups per suburb.

Every score is computed from a versioned, timestamped GTFS snapshot, and the full scoring methodology is published so any number on the site can be independently checked and reproduced.