Data Dictionary¶

Complete reference for all columns across every table in the SUNT dataset.

Conventions

Type: Python/Pandas dtype (str, int, float, datetime, bool).
Nullable ✓: field may contain NaN/NaT values.

Raw Data¶

AVL-lines¶

Column	Type	Description
`route_short_name`	`str`	Bus line identifier (e.g., `"0116"`)
`pt_sequence`	`int`	Stop sequence order along the line
`direction_id`	`int`	`1` = one-way, `0` = return trip
`longitude`	`float`	Longitude of the bus stop
`latitude`	`float`	Latitude of the bus stop
`stop_id`	`str`	Unique stop identifier
`route_long_name`	`str`	Full name/address of the bus stop
`service_code`	`str`	Trip identifier along the line

AVL-vehicles¶

Column	Type	Description
`vehicle`	`str`	Vehicle identifier code
`route_short_name`	`str`	Bus line identifier
`direction_id`	`int`	`1` = one-way, `0` = return
`gps_datetime`	`datetime`	Arrival/departure timestamp at the stop
`longitude`	`float`	Longitude where GPS was recorded
`latitude`	`float`	Latitude where GPS was recorded
`stop_id`	`str`	Stop where the record was captured
`service_code`	`str`	Trip identifier along the line

Timestamp order

For each stop: arrival(N) < departure(N) < arrival(N+1)

AFC¶

Column	Type	Description
`cod_card`	`str`	Anonymized passenger card ID (hash-based)
`afc_datetime`	`datetime`	Timestamp of fare validation
`integration`	`bool`	`True` if part of an integrated transfer
`route_short_name`	`str`	Bus/BRT line identifier
`direction_id`	`str`	`"I"` = one-way, `"V"` = return
`value`	`float`	Fare paid (BRL). `0.0` for free/integrated trips
`vehicle`	`str`	Vehicle where fare was validated

Volume

~35 million records per month.

GTFS Agency¶

Column	Type	Description	Nullable
`agency_id`	`int`	Unique agency identifier
`agency_name`	`str`	Company name
`agency_url`	`str`	Company website	✓
`agency_timezone`	`str`	Timezone (e.g., `"America/Sao_Paulo"`)
`agency_lang`	`str`	Primary language (e.g., `"pt"`)
`agency_phone`	`str`	Contact phone number	✓

GTFS Routes¶

Column	Type	Description
`route_id`	`int`	Unique route identifier
`agency_id`	`int`	Foreign key → GTFS Agency
`route_short_name`	`str`	Short code for the route
`route_long_name`	`str`	Full name (origin × destination)
`route_type`	`int`	`3` = Bus (GTFS standard)

GTFS Trips¶

Column	Type	Description	Nullable
`route_id`	`int`	Foreign key → GTFS Routes
`service_id`	`str`	Groups trips on same operating days
`trip_id`	`str`	Unique trip identifier
`direction_id`	`int`	`0` = one-way, `1` = return
`block_id`	`str`	Block-links consecutive trips	✓
`shape_id`	`str`	Geographic shape of the route	✓

GTFS Stop Times¶

Column	Type	Description
`trip_id`	`str`	Foreign key → GTFS Trips
`arrival_time`	`str`	Scheduled arrival (`HH:MM:SS`)
`departure_time`	`str`	Scheduled departure (`HH:MM:SS`)
`stop_id`	`str`	Foreign key → GTFS Stops
`stop_sequence`	`int`	Order of the stop within the trip
`pickup_type`	`int`	Pickup behavior (`0` = regular)
`drop_off_type`	`int`	Drop-off behavior (`0` = regular)

GTFS Stops¶

Column	Type	Description	Nullable
`stop_id`	`str`	Unique stop identifier. IDs ending `_S` = parent station
`stop_name`	`str`	Human-readable stop name
`latitude`	`float`	Latitude of the stop
`longitude`	`float`	Longitude of the stop
`location_type`	`int`	`0` = stop, `1` = station
`parent_station`	`str`	Parent station `stop_id`, if applicable	✓

LTI — Local Trip Information¶

Column	Type	Description
`route_short_name`	`str`	Bus line identifier
`service_code`	`str`	Trip service identifier
`direction_id`	`str`	`"I"` = one-way, `"V"` = return
`vehicle`	`int`	Vehicle identifier
`start_trip`	`datetime`	Trip start timestamp
`end_trip`	`datetime`	Trip end timestamp
`activity`	`str`	`"Normal"`, `"Leaving the garage"`, or `"Returning to the garage"`

Volume

~700,000 records per month.

Processed Data¶

Boarding¶

Column	Type	Description	Nullable
`tripuserid`	`str`	Composite ID: `{cod_card}_{register_time}`
`type_bus`	`str`	`"bus"`, `"brt"`, or `"subway"`
`user_type`	`str`	Passenger category (e.g., `"driver"`, `"regular"`)
`set`	`str`	Operating company subset ID
`registers`	`int`	Number of validations in this trip
`trip_id`	`str`	`{vehicle}_{route}_{sequence}`
`start_trip`	`datetime`	Trip start timestamp
`end_trip`	`datetime`	Trip end timestamp
`tolerance`	`datetime`	Tolerance window for delay analysis	✓
`integration`	`bool`	Whether this is an integrated transfer
`cod_card`	`str`	Anonymized card ID
`stop_time`	`datetime`	Vehicle arrival at the boarding stop
`register_time`	`datetime`	Fare validation timestamp
`service_code`	`str`	Trip service code
`route_short_name`	`str`	Bus line identifier
`vehicle_afc`	`int`	Vehicle ID from AFC records
`vehicle`	`int`	Vehicle ID from AVL records
`stop_id`	`str`	Stop where boarding occurred
`order`	`float`	Stop sequence order
`direction_id`	`str`	`"I"` or `"V"`
`trip_em`	`float`	Sequential trip number for this vehicle
`dif_boarding`	`float`	Minutes between `register_time` and `stop_time`
`trip`	`str`	`"frt_trip"`, `"mid_trip"`, or `"last_trip"`
`classification`	`str`	`"regular"` or `"irregular"`
`motive`	`str`	Reason for irregular flag (e.g., `"excessive time"`)	✓
`set_nb`	`str`	Neighboring trip company ID	✓
`stop_time_nb`	`datetime`	Neighboring trip stop timestamp	✓
`route_short_name_nb`	`str`	Neighboring trip route	✓
`vehicle_nb`	`int`	Neighboring trip vehicle ID	✓
`stop_id_nb`	`str`	Neighboring trip boarding stop	✓
`diff_nb`	`float`	Minutes to the next boarding event	✓
`motive_pe`	`str`	Quality of neighboring boarding	✓
`target_boarding`	`str`	Final target label

Validity thresholds:

Location	Max `dif_boarding`
Bus station	20 minutes
Regular stop	5 minutes

Alighting¶

Inferred data

All alighting locations are estimated via Trip Chaining. See Methodology → Trip Chaining.

Column	Type	Description	Nullable
`tripuserid`	`str`	Links to the corresponding Boarding row
`stop_time_ali`	`datetime`	Estimated alighting timestamp
`stop_id_ali`	`str`	Stop where alighting is inferred
`order_ali`	`float`	Stop sequence at alighting point
`walk_target`	`str`	Walking behavior: `"regular"` or `"excessive"`
`trip_ali`	`float`	Number of trips associated with this alighting
`walk_dis`	`float`	Estimated walking distance to next stop (km)
`walk_time`	`float`	Estimated walking time (minutes)
`walk_speed`	`float`	Estimated walking speed (km/h)
`diff_de_pe`	`float`	Difference between estimated and observed walking
`wait_time`	`float`	Estimated waiting time at the stop (minutes)
`trip_dis`	`float`	Distance covered during the bus trip (km)
`trip_time`	`float`	Bus trip duration (minutes)
`vel_media`	`float`	Average vehicle speed during the trip (km/h)
`bridge`	`bool`	Whether a walking bridge was used
`bridge_type`	`str`	Type of bridge connection
`bridge_id`	`str`	Bridge identifier	✓
`chain`	`str`	Transport mode sequence (e.g., `"bus-bus"`)
`target_ws`	`str`	Walking speed classification
`target_avs`	`str`	Average speed classification
`target_tt`	`str`	Trip time classification
`target_td`	`str`	Trip distance classification
`target_alighting`	`str`	Final target label

Quality filters:

Filter	Threshold
Max walking distance	1.1 km
Max average vehicle speed	80 km/h
Max trip duration	2 hours

Origin-Destination (OD)¶

Column	Type	Description
`route_short_name`	`str`	Bus line identifier
`register_code`	`int`	Service registration code
`direction_id`	`str`	`"I"` or `"V"`
`pt_sequence`	`int`	Stop sequence number in the trip
`stop_id`	`int`	Stop identifier
`vehicle`	`int`	Vehicle identifier
`trip_number`	`int`	Sequential trip number
`trip_id`	`str`	`{vehicle}_{route}_{sequence}`
`start_trip`	`datetime`	Trip start timestamp
`end_trip`	`datetime`	Trip end timestamp
`stop_time`	`datetime`	Vehicle arrival at this stop
`n_boardings`	`float`	Passengers boarding at this stop
`n_alighting`	`float`	Passengers alighting at this stop
`lag_loading`	`int`	Passenger load before vehicle arrives
`balance`	`int`	Load after alightings: `lag_loading − n_alighting`
`loading`	`int`	Load after boardings: `balance + n_boardings`

First stop

At pt_sequence = 1: lag_loading = balance = n_alighting = 0.

Graph Data¶

Edge Features¶

Column	Type	Description
`src`	`int`	Origin node (stop/station ID)
`dst`	`int`	Destination node (stop/station ID)
`distance`	`float`	Distance between stops (km)
`src_lat`	`float`	Latitude of origin stop
`dst_lat`	`float`	Latitude of destination stop
`src_lon`	`float`	Longitude of origin stop
`dst_lon`	`float`	Longitude of destination stop
`average_speed`	`float`	Mean vehicle speed on this segment (km/h)
`trip_time`	`int`	Mean travel time on this segment (minutes)
`loading`	`int`	Mean passenger load on this segment

Node Features (Static)¶

Column	Type	Description
`node`	`int`	Stop/station identifier
`loading`	`float`	Average passengers passing through
`n_alighting`	`float`	Average alightings per vehicle visit
`n_routes`	`float`	Average distinct routes serving this node
`n_boarding`	`float`	Average boardings per vehicle visit
`n_trips`	`float`	Average trips per day
`n_vehicles`	`float`	Average distinct vehicles per day
`average_speed`	`float`	Average vehicle speed on arrival (km/h)

Node Features (Temporal)¶

Aggregated into 5-minute intervals.

Column	Type	Description
`interval`	`datetime`	Start of the 5-minute observation window
`node`	`int`	Stop/station identifier
`loading`	`float`	Passengers at the node in this interval
`n_boarding`	`float`	Boardings in this interval
`n_alighting`	`float`	Alightings in this interval
`n_vehicles`	`float`	Distinct vehicles in this interval
`n_routes`	`float`	Distinct routes in this interval
`n_trips`	`float`	Distinct trips in this interval
`average_speed`	`float`	Average vehicle speed (km/h)