GTFS2GMNS

The open-source Python codes (GTFS2GMNS) is released to facilitate researchers and planners to construct the multi-modal transit networks easily from generic General Transit Feed Specification (GTFS) to the network modeling format in General Modeling Network Specification (GMNS). The converted physical and service networks in GMNS format are more convenient for network modeling tasks such as transit network routing, traffic flow assignment, simulation and service network optimization.

Your comments will be valuable for code review and improvement. Please feel free to add your comments to our Google document of GTFS2GMNS Users' Guide.

Getting Started

Download GTFS Data

On TransitFeed homepage, users can browse and download official GTFS feeds from around the world. Make sure that the following files are present, so that we can proceed.

• stop.txt
• route.txt
• trip.txt
• stop_times.txt
• agency.txt

GTFS2GMNS can handle the transit data from several agencies. Users need to configure different sub-files in the same directory. Under the test/GTFS folder, a subfolder Pheonix with its owm GTFS data is set up.

Convert GTFS Data into GMNS Format

if __name__ == '__main__':
    global period_start_time
    global period_end_time
    input_gtfs_path = 'GTFS'
    output_gmns_path = '.'
    time_period_id = 1
    time_period = '1200_1300'
    period_start_time, period_end_time = _hhmm_to_minutes(time_period)

    gtfs2gmns(input_gtfs_path, output_gmns_path)

The input parameter input_gtfs_path is the path of GTFS data, and the parameter output_gmns_path is the path of output GMNS files. Users can customize the parameter time_period, such as 12:00 to 13:00.

The output files include node.csv and link.csv.

Main Steps

Read GTFS data

Step 1.1: Read routes.txt

• route_id, route_long_name, route_short_name, route_url, route_type

Step 1.2: Read stop.txt

• stop_id, stop_lat, stop_lon, direction, location_type, position, stop_code, stop_name, zone_id

Step 1.3: Read trips.txt

• trip_id, route_id, service_id, block_id, direction_id, shape_id, trip_type
• and create the directed_route_id by combining route_id and direction_id

Step 1.4: Read stop_times.txt

• trip_id, stop_id, arrival_time, deaprture_time, stop_sequence

• create directed_route_stop_id by combining directed_route_id and stop_id through the trip_id

  Note: the function needs to skip this record if trip_id is not defined, and link the virtual stop id with corresponding physical stop id.

• fetch the geometry of the direction_route_stop_id

• return the arrival_time for every stop

Building service network

Step 2.1 Create physical nodes

• physical node is the original stop in standard GTFS

Step 2.2 Create directed route stop vertexes

• add route stop vertexes. the node_id of route stop nodes starts from 100001

  Note: the route stop vertex the programing create nearby the corresponding physical node, to make some offset.

• add entrance link from physical node to route stop node

• add exit link from route stop node to physical node. As they both connect to the physical nodes, the in-station transfer process can be also implemented

Step 2.3 Create physical arcs

• add physical links between each physical node pair of each trip

Step 2.4 Create service arcs

• add service links between each route stop pair of each trip

Visualization

You can visualize generated networks using NeXTA or QGIS.

Quick Tutorial

Introducing Functional Tools within GTFS2GMNS

GTFS2GMNS is a Python package that serves as a class-based instance, specifically designed for reading, converting, analyzing, and visualizing GTFS data. The converted physical and service networks in GMNS format offer enhanced convenience for a variety of networkAg modeling tasks, including transit network routing, traffic flow assignment, simulation, and service network optimization.

Input for class GTFS2GMNS

• gtfs_input_dir : str, the dir store GTFS data. GTFS2GMNS is capable of reading multiple GTFS data sets.
• time_period: str, the time period sprcified (for data selection), default is "07:00:00_08:00:00"
• date_period: list, user can specified exact data or dates for selection
• gtfs_output_dir: str, the output folder to save data. defalut is ""
• isSaveToCSV: bool, whether to save gmns node and link to local machine, default is True

Code Example

Loading gtfs data

from gtfs2gmns import GTFS2GMNS

if __name__ == "__main__":
    gtfs_input_dir = r"Your-Path-Folder-To-GTFS-Data"

    # Explain: GMNS2GMNS is capable of reading multiple GTFS data sets
    """
	--root folder
	    -- subfolder (GTFS data of agency 1)
	    -- subfolder (GTFS data of agency 2)
	    -- subfolder (GTFS data of agency 3)
	    -- ...
	then, assign gtfs_input_foler = root folder
    """

    time_period = "00:00:00_23:59:59"
    date_period = []

    gg = GTFS2GMNS(gtfs_input_dir, time_period, date_period, gtfs_output_dir="", isSaveToCSV=False)

Generate access line between zones to nodes

import gtfs2gmns as gg

path_zone = "Path to zone.csv"    # please make sure you have zone_id, x_coord, y_coord in columns
path_node = "Path to node.csv"    # please make sure you have node_id, x_coord, y_coord in columns

radius = 100 # unit in meters
k_closest = 0 # if 0, generate all accessible links within radius. if 1, closest link within the radius...

access_links = gg.generate_access_lins(path_zone, path_node, radius, k_closest)

access_links.to_csv("access_link.csv", index=False)

Functions and Attributes

func_type	func_name	Python example	Input	Output	Remark
read-show	agency	gg.agency	NA	DataFrame	This attribute load and return agency data from source folder
	calendar	gg.calendar
	calendar_dates
	fare_attributes
	fare_rules
	feed_info
	frequencies
	routes
	shapes
	stops
	stop_times
	trips
	transfers
	timepoints
	timepoint_times
	trip_routes
	stops_freq
	routes_freq
	rute_segments
	route_segment_speed
	vis_stops_freq
analysis	vis_routes_fres
	vis_route_segment_speed
	vis_route_segment_runtime
	vis_route_stop_speed_heatmap
	vis_spacetime_trajectory
	equity_alanysis
	accessibility_analysis
	load_gtfs
	gen_gmns_node_link
visualization

Upcoming Features

• Output service and trace files.
• Set the time period and add vdf_fftt and vdf_freq fields in link files.
• Add Visualization functions
  • Stops
  • Routes
  • ...

Citing gtfs2gmns

If you use gtfs2gmns in your research please use the following BibTeX entry:

@misc{gtfs2gmns,
  author =       {Xiangyong Luo, Fang Tang, Han Wang, Xuesong Zhou},
  title =        {gtfs2gmns - A class-based instance designed for reading, converting, analyzing, and visualizing GTFS data.},
  howpublished = {Github},
  year =         {2023},
  url =          {https://github.com/xyluo25/gtfs2gmns}
}

@misc{gtfs2gmns,
  author =       {Xiangyong Luo, Fang Tang, Han Wang, Xuesong Zhou},
  title =        {gtfs2gmns - A class-based instance designed for reading, converting, analyzing, and visualizing GTFS data.},
  howpublished = {Github},
  year =         {2023},
  url =          {https://github.com/xyluo25/gtfs2gmns}
}