1. Route operating profiles
Default scope: bus routes identified by the static feed. Unassigned vehicles are excluded because their garage or layover activity can dominate the results. Time categories use the mean of reported speeds at both ends of accepted intervals. Stationary ≤1 km/h; slow moving >1–10 km/h. These are estimated vehicle-minutes, not passenger time.
Chart: top 12 routes by total stationary + slow vehicle-time. Bar length shows total accepted vehicle-time.
| Route | Vehicles | Vehicle-min | Sampled km | Ground speed km/h | Stationary % | Slow moving % | Unknown speed % |
|---|
Click a route to filter. Ground speed = summed straight-line segment distance / accepted interval time; it can underestimate road distance. Percentages use all accepted time. Route profiles combine branches unless direction is filtered.
Changes during the window
15-minute bins, assigned by interval midpoint. Fleet composition can change between bins.
2. Repeated slow locations
Locations are screening candidates for field investigation. They are ranked by total stationary + slow vehicle-minutes in approximately 250 m cells, with at least three different vehicles.
| Cell / nearby stop | Vehicles | Low-speed min | Stationary min | Total min | Low-speed % |
|---|
The map shows the top 150 eligible cells; the export includes all. Cells combine nearby roads, directions and stops within the selected filters. Interval midpoints are approximate locations, not map-matched road segments.
3. Travel time between observed stop areas
Each sample connects consecutive detected stop-area passages on the same uninterrupted trip. Passages use the closest recorded point within 60 m. Some intermediate stops are missed. Terminal-linked comparisons are excluded by default because waiting at endpoints can dominate travel time. Click a row to locate its two endpoints.
| Route / direction | From → to | Samples / vehicles | P20 min | Median min | P90 min | P90 / median | Median schedule difference min | Total excess vs P20 min |
|---|
Observed sample travel times
How to interpret the comparison
P20 is the 20th percentile of observed times for the same route pattern and stop pair in the selected window. It is an internal benchmark, not a measured free-flow time.
P90 / median describes the spread in this small sample. “Excess” sums positive differences above P20; it is not causal traffic delay. Schedule difference compares passage-to-passage time with scheduled arrival-to-arrival time.
Short links are sensitive to the roughly 30-second position updates and the 60 m detection radius. Differences of a few seconds are not meaningful here.
4. Approximate schedule deviation
Screening only: closest observation within 30 m of a scheduled timepoint, minus its scheduled arrival time. Positive means later than the listed arrival. This is not a confirmed arrival or official on-time performance metric.
| Route | Passages | Vehicles | P10 offset min | Median offset min | P90 offset min | |Offset| >60 min |
|---|
Trip ID and route ID must match the static feed active September 22. The static file was downloaded September 23; it does not establish the schedule version published the day before. Long terminal stays, nearby roads, trip assignment and timestamp alignment may bias offsets. Extreme values are retained and flagged, not silently removed.
5. Spacing between observed buses
Time gaps between detected passages by different vehicles at the same stop and identical route pattern. Missing detections can inflate gaps. This is a headway proxy, not proof of missing service or bunching.
| Route / direction | Stop / pattern | Gaps / vehicles | Median min | P90 min | CV | Short-gap % |
|---|
Time–stop diagram
Time runs left to right; vertical position is stop order along the chosen pattern. Each color is a vehicle. Dots are detected stop-area passages; connecting lines are only visual guides. Lines break at trip-episode boundaries. Close lines suggest a spacing question to investigate, not a confirmed bunching event.
6. Stops and vehicle trip fragments
Recorded presence near stops
| Stop | Passages / vehicles | Median recorded span s | P90 span s | Median nearest distance m |
|---|
A span is the time from first to last consecutive recorded observation in a 60 m stop area. One point has span zero, which means duration unresolved. These spans can include queues or layovers and are not passenger dwell times.
Uninterrupted observed trip fragments
| Vehicle | Route / direction | Trip | From | To | Observed min | Sampled km | Terminal proximity span |
|---|
7. Transit-priority sensitivity scenarios
These are adjustable arithmetic scenarios, not forecasts or benefit estimates. Evaluate each separately; their time bases can overlap.
A · Reduce part of benchmark excess
Eligible stop-pair excess × assumed attribution × assumed reduction. Examples to investigate include queue bypass or signal priority. The observations do not supply either assumption.
B · Reduce part of stationary time
All selected estimated stationary seconds × eligible share × reduction. Passenger boarding, safety stops and layover requirements cannot simply be removed.
Candidate interventions and the evidence still needed
| Intervention | Screen with this report | Before recommending it |
|---|---|---|
| Transit signal priority | Repeated slow locations; link variability | Intersection geometry, signal phases, queue and arrival data |
| Bus lane / queue jump | Repeated low-speed exposure across vehicles | Road lanes, traffic volumes, mixed-traffic comparison, feasibility |
| Stop or boarding changes | Repeated near-stop presence spans | Door events, passenger counts, accessibility and stop spacing |
| Headway-based control | Repeated short and long passage gaps | Complete arrivals, planned headways, dispatch and holding records |
| Schedule / recovery review | Schedule-offset distribution; trip fragments | Verified schedule version, full trips across multiple days |
8. Findings to discuss
What this collection cannot establish
Passenger delay, passenger-weighted benefit, intersection red-light delay, actual signal-priority activations, reliable end-to-end route times, peak-hour performance, day-to-day reliability and causal congestion effects require additional observations. The dataset can include buses and streetcars; vehicle mode is not inferred from its ID.
Methods and quality checks
Geofencing uses a local distance approximation at Toronto latitude. Repeated stop IDs within a trip are excluded to avoid ambiguous loop-stop matching. One closest passage per stop per uninterrupted trip fragment is retained. Travel comparisons are grouped by route, direction, shape and ordered stop pattern; different patterns are not pooled. Small samples are descriptive, not independent statistical replicates.
For schedule interpretation, see the official GTFS Schedule reference. A realtime stop ID can mean an upcoming or current stop; see Vehicle Positions documentation. This analysis uses geometric proximity rather than treating the reported stop ID as an arrival event.
Official static source: TTC Surface Routes and Schedules for BusTime. Contains information licensed under the Open Government Licence – Toronto. Street tiles © OpenStreetMap contributors.