Research Methodology
Methodology version v1.0. All thresholds and index weights are configurable in the methodology settings and versioned with every analysis run.
The Corridor Planning Principle
Each route is timed independently — arrivals on the shared street may collide or leave long gaps.
→ coordinated combined service
→ shorter effective corridor headway
→ reduced clustering
→ reduced passenger waiting
Scheduled Bunching Classification (vv1.0)
For consecutive combined arrivals along the same shared corridor:
| Gap between consecutive arrivals | Classification |
|---|---|
| ≤ 2 min | Severe bunching |
| >2 and ≤ 5 min | Moderate bunching |
| >5 and ≤ 15 min | Coordinated service |
| >15 and ≤ 30 min | Moderate service gap |
| > 30 min | Long service gap |
Research Metrics
Bunching Index OKEN research metric
0–100, where 0 = highly coordinated and 100 = severely irregular. Weighted combination of the share of gaps ≤ 5 min (w=0.35), coefficient of variation (w=0.25), long-gap rate (w=0.2) and deviation from the corridor target headway (w=0.2). Weights are editable in methodology settings.
Research metric developed by OKEN Transit Research Lab. Not an official MassDOT or transit-agency metric.
Corridor Regularity Score
Complementary 0–100 score (higher = better) derived from headway consistency, reduced extreme gaps, reduced clustering and proximity to the target headway. Bands: Excellent ≥85 · Good ≥70 · Moderate ≥50 · Poor ≥30 · Critical <30.
What is a corridor?
A corridor is a street segment shared by two or more bus routes. Along the shared segment, passengers can use any of the overlapping routes, so the service they experience is the combined schedule of all routes — not any single route's schedule.
What is route overlap?
Two routes overlap when they serve the same stops along the same alignment. Overlap is verified from GTFS stops.txt, stop_times.txt and shapes.txt — never assumed from route maps or route numbers.
What is headway?
Headway is the time gap between consecutive vehicle arrivals at a stop or control point. On a shared corridor, the combined headway is the gap between consecutive arrivals of any corridor route.
What is bus bunching?
Bus bunching occurs when vehicles arrive close together followed by long gaps, instead of at even intervals. Passengers experience long waits and crowded vehicles even when the average headway looks acceptable.
What is scheduled bunching?
Scheduled bunching is clustering built into the timetable itself: the GTFS schedule places arrivals minutes apart, followed by long scheduled gaps. This study measures scheduled bunching only.
What is observed bunching?
Observed bunching is measured from vehicle-location data (GTFS-Realtime, AVL, APC). It captures what actually happened, including traffic and operations effects. A future module will compare scheduled vs observed arrivals.
Why combined corridor headway matters
Where routes overlap, coordinating their schedules can halve effective waiting time without adding a single bus. Conversely, uncoordinated schedules on overlapping routes can produce simultaneous departures and long gaps.
Why average headway alone can be misleading
Two buses an hour can mean one every 30 minutes — or two arriving together every 60. The average is identical; the passenger experience is not. That is why this lab reports distributions, variance and bunching rates alongside averages.
Coefficient of Variation (CV)
CV is the standard deviation of headways divided by the mean. A CV near 0 means perfectly even spacing; higher values mean irregular service. CV is scale-free, so corridors with different frequencies can be compared fairly.
Waiting-time implications
With random passenger arrivals, expected waiting time is E[W] = Σh² / (2Σh). Irregular headways increase expected waiting time even when the average headway is unchanged — this is the core passenger harm from bunching.
Long service gaps
Gaps beyond 30 minutes are treated as long service gaps. On a corridor intended to be high-frequency, even a few long gaps per period severely degrade usability.
Schedule coordination
Coordination means offsetting overlapping routes' timetables so combined arrivals approach even spacing at the corridor's target headway — using existing resources before adding service.
Analysis date selection
The three analysis dates—June 26, July 31, and August 21, 2026—were selected as comparable Friday service days to support a like-for-like longitudinal assessment of scheduled corridor headway regularity, service gaps, and potential bunching. GTFS calendar and calendar-exception records are validated for each date before comparison. The WRTA longitudinal comparison runs June 26, 2026 (baseline) → July 31, 2026 (midpoint) → August 21, 2026 (latest observation).
GTFS methodology
Analysis uses static GTFS feeds. Service on each analysis date is resolved from calendar.txt and calendar_dates.txt (including exception types and post-midnight times beyond 24:00:00). Headways are measured at a consistent control point per corridor.
Limitations
GTFS represents planned service. Scheduled clustering is not proof of actual bunching. Traffic, boarding, operator behavior, incidents and dispatch interventions change real operations. Observed analysis requires AVL/APC/GTFS-Realtime data.
Coming Soon — Post-Corridor Scheduling Ridership Modeling
Planned tools: PTV Mobility, GIS, GTFS, ridership data, OD demand, population, employment and accessibility. Planned outputs: existing and projected ridership, passenger-km, boardings, alightings, vehicle occupancy, waiting time, transfers, accessibility and scenario comparison. No ridership projections are published in Version 1.
Coming Soon — Sustainability Analysis
Planned module: transit attractiveness, potential mode shift, vehicle travel avoided, fuel and emissions implications, accessibility improvements and equity indicators. Measured results will always be distinguished from modeled estimates.
Coming Soon — Real-Time Module
Architecture prepared for GTFS-Realtime Trip Updates, Vehicle Positions, Service Alerts, AVL and APC — enabling scheduled vs actual arrival comparison: actual headway, schedule deviation, actual bunching, recovery, dwell time and reliability. No simulated live data is shown.