A GTFS frequency diagnostic tool built on Brampton Transit's real feed that classifies the frequent network, flags service cliffs, and lets you simulate adding trips to make a weak route frequent.
The frequent network classifier marks a route frequent when its worst hourly median headway across 7am to 6pm is 15 minutes or better. The cliff detector finds the first daytime hour where headway doubles and flags that as a service cliff. The what-if simulator lets you add trips to any hour and watch the classification update in real time. A per-route auto-diagnosis line ties those three metrics into a plain-English verdict at the top of each route panel.
The analysis runs against a single representative weekday resolved from the feed's service calendar: a Wednesday near the middle of the feed's date range, verified to have a healthy trip count. Frequent is defined as a worst-case hourly median headway of 15 minutes or better across hours 7 through 18. The cliff threshold is tuned to fire only when a daytime hour's headway crosses out of the frequent range, which excludes routine late-night wind-down and isolates genuine mid-service drops. Full details are in pipeline/METHODOLOGY.md.
Python and Polars for the pipeline, React and Vite for the frontend, Recharts and Leaflet for the visuals, deployed on Vercel.
data/raw: raw Brampton Transit GTFS feedpipeline/: Polars script that generates the JSON analysis filesweb/: React appdocs/: screenshotspipeline/METHODOLOGY.md: full methodology