Coverage and interpretation
A feed is not necessarily one agency. Identifiers and stop counts are local to this feed; duplicates across providers are not merged.
Service-date timeline
Hover for values; click to inspect a service date.
Departures by service-day hour
Trips with unavailable start times are omitted. Agency-local time. Hours ≥24 remain attached to the originating service date. Frequency-based counts may be nominal.
Endpoint duration distribution
First departure to final arrival, including intermediate dwell; excludes missing or negative endpoint durations. Not observed travel time.
Weekday and month comparisons · selected plot range
| Weekday | Days | Mean/day | Min–max |
|---|
| Month | Covered days | Mean/day | Occurrences |
|---|
Zero-service dates within the selected interval are included. Partial months are not extrapolated; changing weekday mix can affect means.
Largest changes versus seven days earlier · selected plot range
| Service date | Earlier date | Earlier count | Count | Change | Feed-wide exception effect* |
|---|
*Count after calendar exceptions minus the base calendar count, across the whole feed. A change in this snapshot is not evidence of an observed cancellation. Dates outside the analysis interval are unknown, not zero.
Stop geography
All valid coordinates, including stations and other stop-location types. Geographic scatter plot without a basemap; not moving vehicles.
Route types and accessibility
Wheelchair code 0/blank is unspecified and may inherit from a parent; 1 declares accessibility; 2 declares no accessibility. These attributes are not a field audit.
Route comparison · selected date
| Route | Name | Mode | Distinct stops | Departures* | Nominal subset | Median min | P90 min | First–last |
|---|
*Fixed-time and exact frequency departures plus nominal slots for non-exact frequency service. Directions are combined. Stops are distinct references across all templates, not just this date. First–last is first departure through final arrival, potentially with gaps.
Data quality
New targeted structural checks
| Check | Findings |
|---|
These checks do not replace a full GTFS validator. Blank intermediate times can be permitted; shape absence and coordinate requirements depend on the GTFS feature used. Findings can overlap and must not be summed as unique bad records.
Methods, limits and sources
CSV spaces immediately following separators are tolerated during parsing; source files remain unchanged. Every date in the interval is evaluated using calendar.txt weekday flags and date bounds, then calendar_dates.txt additions/removals. Complete feed_info date bounds take precedence. If missing, bounds come from weekly-active calendars and added service dates referenced by trips; all-zero placeholder calendars do not set the inferred horizon. This differs from the supplied validator's service-window fields when those include broad placeholder dates.
Trip templates are rows in trips.txt, not daily departures or unique vehicles. Frequency templates replace, rather than supplement, their single base trip. For a frequency window, starts are generated from start_time (inclusive) to end_time (exclusive) at headway_secs intervals. exact_times=1 contributes exact scheduled departures; exact_times=0 or blank contributes nominal analytical slots only. Their combined count is a departure equivalent; nominal trip-hours, calls, hourly placement and percentiles are estimates under this expansion, not guaranteed schedules.
Durations use endpoint differences. Percentiles use linear interpolation weighted by expanded occurrences. Trip-hours sum these durations and exclude inter-trip layovers and deadheading. No actual speed, punctuality, ridership or fleet size is inferred. The database's GeoPackage has simplified route geometry; use the original GTFS shapes for detailed planned paths.