generated: '2026-07-28'
method: derived
source: gtfs/amtrak-gtfs.zip (feed_version 20260727), columns and counts read from the archive
note: >-
  Derived from the harvested feed itself, not from documentation - Amtrak
  publishes no object reference. Column lists are the exact header rows of each
  member file; row counts exclude the header. Semantics follow the GTFS Schedule
  reference at https://gtfs.org/documentation/schedule/reference/.
entities:
- name: feed_info
  file: feed_info.txt
  rows: 1
  primary_key: null
  fields: [feed_publisher_name, feed_publisher_url, feed_lang, default_lang, feed_start_date, feed_end_date, feed_version, feed_contact_email, feed_contact_url]
  values:
    feed_publisher_name: Amtrak
    feed_publisher_url: http://www.amtrak.com
    feed_lang: en
    feed_version: '20260727'
    feed_contact_email: DL_DTGTFSsupport@Amtrak.com
- name: agency
  file: agency.txt
  rows: 20
  primary_key: agency_id
  fields: [agency_id, agency_name, agency_url, agency_timezone, agency_lang]
  note: >-
    Amtrak itself is agency_id 51. The other 19 are Thruway bus and charter
    operators (Roadrunner Shuttle, Arrow Trailways, Coach Usa, FLIX, Beeline
    Express, Vegas Airporter, Stagecoach Express, The Lift, Executive
    Transportation, Alvand Transportation, Route 66 Express, Sunway Charters,
    Amtrak Chartered Vehicle/Motorcoach, Shore Line East, MARC) plus Via Rail
    Canada (agency_id 157), the partner on the jointly operated Maple Leaf.
- name: routes
  file: routes.txt
  rows: 61
  primary_key: route_id
  fields: [route_id, agency_id, route_short_name, route_long_name, route_type, route_url, route_color, route_text_color]
  breakdown:
    route_type_2_rail: 49
    route_type_3_bus: 12
    distinct_route_long_name: 48
  note: >-
    route_short_name is empty throughout; route identity is carried entirely in
    route_long_name (Acela, Auto Train, California Zephyr, Coast Starlight,
    Empire Builder, Crescent, Cardinal, Amtrak Cascades, Capitol Corridor,
    Amtrak Hartford Line, Downeaster, Borealis, Berkshire Flyer, Floridian,
    Mardi Gras Service, Amtrak Thruway Connecting Service and 32 more).
- name: stops
  file: stops.txt
  rows: 646
  primary_key: stop_id
  fields: [stop_id, stop_code, stop_name, stop_url, stop_timezone, stop_lat, stop_lon]
  note: >-
    stop_id and stop_code are both the Amtrak three-letter station code and are
    identical in every row (CHI, ABE, ABQ, ADM). stop_url is the deep link
    https://www.amtrak.com/stations/<lowercased code>.
- name: calendar
  file: calendar.txt
  rows: 403
  primary_key: service_id
  fields: [service_id, monday, tuesday, wednesday, thursday, friday, saturday, sunday, start_date, end_date]
  span: {start: '20260726', end: '20270726'}
- name: trips
  file: trips.txt
  rows: 2948
  primary_key: trip_id
  fields: [route_id, service_id, trip_id, trip_short_name, direction_id, shape_id, trip_headsign]
  note: trip_short_name is the public Amtrak train number - the key downstream tracking projects join on.
- name: stop_times
  file: stop_times.txt
  rows: 37862
  primary_key: [trip_id, stop_sequence]
  fields: [trip_id, arrival_time, departure_time, stop_id, stop_sequence, pickup_type, drop_off_type, timepoint]
- name: shapes
  file: shapes.txt
  rows: 373236
  primary_key: [shape_id, shape_pt_sequence]
  fields: [shape_id, shape_pt_lat, shape_pt_lon, shape_pt_sequence]
relationships:
- from: routes
  to: agency
  type: belongs_to
  via: agency_id
- from: trips
  to: routes
  type: belongs_to
  via: route_id
- from: trips
  to: calendar
  type: belongs_to
  via: service_id
- from: trips
  to: shapes
  type: belongs_to
  via: shape_id
- from: routes
  to: trips
  type: has_many
  via: route_id
- from: trips
  to: stop_times
  type: has_many
  via: trip_id
- from: stop_times
  to: trips
  type: belongs_to
  via: trip_id
- from: stop_times
  to: stops
  type: belongs_to
  via: stop_id
- from: stops
  to: stop_times
  type: has_many
  via: stop_id
- from: calendar
  to: trips
  type: has_many
  via: service_id
- from: shapes
  to: trips
  type: has_many
  via: shape_id
absent_entities:
- calendar_dates (no service exceptions published)
- fare_attributes / fare_rules / fare_products (no fare data at any GTFS generation)
- frequencies, transfers, pathways, levels, translations, attributions
- any reservation, order, passenger, payment or loyalty entity - the transactional data model is unpublished