No matching stops found. Try a different name or 5-digit code.
No services found for this stop.
Not enough data yet — keep the app running to build history.
The AI uses a Gradient Boosting Regressor to predict how many seconds each bus will arrive earlier or later than the LTA estimate. The adjusted time is shown in blue.
The background collector tracks each bus across API polls. When it disappears it has arrived. delay = last_estimate − first_estimate. The model retrains hourly.
Day one uses 5 000 synthetic rows. Real data is blended in with 3× weight as it accumulates.
Tap a card to load live arrivals. Saved in your browser.
No favourites yet — search for a stop and tap ☆ to save it.
Live view of what's stored — use this to verify your database is connected.
| Stop | Service | First Seen | Delay (s) |
|---|
| Stop | Service | Time | Wait (s) | Delay (s) | Load | Peak |
|---|