facct_2022_dataset

The dataset for our FAccT 2022 Paper "Equitable Public Bus Network Optimization for Social Good: A Case Study of Singapore" is available on the following google drive link

Please cite our paper if using our dataset:

@inproceedings{Tedjopurnomo2022,
    title={Equitable Public Bus Network Optimization for Social Good: A Case Study of Singapore},
    author={Tedjopurnomo, David and Bao, Zhifeng and Choudhury, Farhana and Luo, Hui and Qin, A. K.},
    year = {2022},
    booktitle = {Proceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency},
    series = {FAccT '22}
}

There are nine pickle (.pkl) files in this repository, the details of which are given below

all_areas.pkl

Click to expand

This file contains the data of Singapore's planning areas. There are 55 Singapore planning areas, however, as 5 of them do not have any bus stops data, we only record the data of 50 planning areas. This is a nested dictionary. To access the level-two dictionary containing data of one area, query the dictionary by providing the area name `String` as the key in uppercase (e.g. "PASIR RIS")

The inner dictionary contains the following key-value pairs:

Key	Value
Shape	String that says Polygon (Unused)
Planning Area Name	String of the area name; same as the level-one dictionary key
Planning Area Code	String for the shorthand code of area name
Central Area Indicator	String that indicates the central area this planning area is in
Region Name	String that shows in which of Singapore's five regions this area is located in
Region Code	String shorthand code for the Region Name
INC_CRC	String possibly denoting unique code for the area (Unused)
FMEL_UPD_D	String original upload date of the data (Unused)
X_ADDR	String possibly storing the x-axis location of the area (Unused)
Y_ADDR	String possibly storing the y-axis location of the area (Unused)
SHAPE_Length	String denoting the length of the area (Unused)
SHAPE_Area	String denoting the size of the area in kilometers
Boundaries	shapely.geometry.polygon.Polygon containing the area's boundaries
Bus Stops	List containing this area's bus stops, each denoted with a String ID
Race	List containing tuples of String and integer for the race name and population count of that race respectively
Sex	List containing tuples of String and integer for the sex name and population count of that sex respectively
Qualification	List containing tuples of String and integer for the qualification name and population count of that qualification respectively
Income	List containing tuples of List and integer for the income ranges and population count within that income range respectively
Stop to Area	A two-level nested dict containing the stop-to-area efficiency data. The key to the first level dictionary is the String of the stop ID within the area and the key to the second level is the String of the area name
Stop to Stop	A two-level nested dict containing the stop-to-stop efficiency data. The key to the first level dictionary is the String of the stop ID within the area and the key to the second level is the String of another stop ID
Car Distance	A two-level nested dict containing the car distance data. The key to the first level dictionary is the String of the stop ID within the area and the key to the second level is the String of another stop ID, The unit is in meters.
Bus Distance	A two-level nested dict containing the bus distance data, which includes the waiting distance and transfer penalty. The key to the first level dictionary is the String of the stop ID within the area and the key to the second level is the String of another stop ID, The unit is in meters.
Stop Coverage	dict with a String as a key and a float as the value. The key denotes a bus stop ID and the float denotes the coverage area of that bus stop. Specifically, each bus stop is assigned a non-overlapping 400 square meters radius for its coverage. The float here stores a bus stop's fraction of area coverage over all bus stop's coverage in that area. For instance, if an area has two bus stops with non-overlapping area total of 800 square meters, each bus stop will have a fraction of 0.5
Area to Area	dict with a String as a key and Decimal as the value. The key denotes another area's name and the value denotes this area's efficiency to that area
Disadvantaged	dict with a String as a key and bool as the value. The key denotes a demographic factor and the value denotes if this area is a disadvantaged area for that demographic factor
Disadvantaged Percentage	dict with a String as a key and float as the value. The key denotes a demographic factor and the float denotes the fraction of disadvantaged population for that demographic factor
Average Connectivity	Decimal denoting this area's overall connectivity/efficiency

all_nodes.pkl

Click to expand

This file contains the data of the nodes of Singapore's car graph. This is a nested dictionary. To access the level-two dictionary containing data of one node, query the dictionary by providing the node ID `String`. There are two types of nodes: bus stops and road junctions. Bus stop IDs are shorter with about 4-5 characters while road junction IDs are longer with about 9-10 characters.

The inner dictionary contains the following key-value pairs:

Key	Value
ID	String of the node ID, same as the level-one dictionary key
coordinates	tuple of floats for the latitude and longitude coordinates of the node

all_edges.pkl

Click to expand

This file contains the data of the edges of Singapore's car graph. This is a nested dictionary. To access the level-two dictionary containing data of one edge, query the dictionary by providing the edge ID `String`.

The edges have underscore components which denotes the original edge that has been split on the bus stop contained in that edge. For instance, edges 39449_1 and 39449_2 are splits of the original edge 39449, where the split is done on bus stop 96449. The original edge 39449 starts on junction with ID 5060088362 and ends on junction with ID 2327165732. After the split, edge 39449_1 starts on junction with ID 5060088362 and ends on bus stop 96449, while edge 39449_2 starts on bus stop 96449 and ends on junction with ID 2327165732

The inner dictionary contains the following key-value pairs:

Key	Value
Bus Stops	An empty list (Unused)
Coordinates	list of tuples of floats containing the latitude-longitude pair that constitutes the edge
ID	String of the edge ID, same as the level-one dictionary key
Length	float for the length of the edge in meters
u	String of node ID which is the origin of the edge. The node ID is the same as in all_nodes.pkl
v	String of node ID which is the destination of the edge. The node ID is the same as in all_nodes.pkl

all_stop_mapmatched.pkl

Click to expand

This file contains the data of Singapore's bus stops after the mapmatching process. This is a nested dictionary. To access the level-two dictionary containing data of one stop, query the dictionary by providing the bus stop ID `String`.

The inner dictionary contains the following key-value pairs:

Key	Value
Linestring Old	String containing the linestring of the old (before mapmatching) coordinates
Name	String of the bus stop name
Edge ID	String of the edge ID the bus stop is on. This ID represents the original (i.e. pre-split) edge ID. Refer to the documentation for all_edges.pkl above for details
Road Name	String of the road name the bus stop is on. This refers to the Edge ID above
Coordinates Old	tuple of floats of latitude-longitude pairs where the bus stop is located. This is before the mapmatching process
Coordinates New	tuple of floats of latitude-longitude pairs where the bus stop is located. This is after the mapmatching process
Valid Start	bool denoting whether or not this is a terminal bus stop. In Singapore's case, it is whether or not this bus stop is the start or ending point of at least one bus route

bus_routes.pkl

Click to expand

This file contains the data of Singapore's bus routes. This is a nested dictionary. To access the level-two dictionary containing data of one route, query the dictionary by providing the bus route ID `String`.

The inner dictionary contains the following key-value pairs:

Key	Value
seq_no	list of int starting from 1 to the number of bus stops in the route (Unused)
stop_id	list of Strings representing the bus stop IDs of the bus route
dist	list of Decimals representing the distance from the n-th bus stop in the route to the start of the route
wd_first	String representing the bus route's operation start time in the HHMM 24-hour format without any leading zero for the hour
wd_last	String representing the bus route's operation end time in the HHMM 24-hour format without any leading zero for the hour
frequency	list of tuples where each tuple contains two datetime.time objects and a float. The two datetime.time objects denote a time period and the float denotes the bus frequency/waiting time for that period
startend	list of two datetime.time objects containing the bus route's operation start and end time
route_type	String denoting the type of the route. For the Singapore data, the only value will be bus

bus_dists.pkl and car_dists.pkl

These files contain the data of Singapore's bus and car distances in meters. This is a nested dictionary where the key for the first and second level dictionary are both bus stop ID Strings. The value is the bus or car distance in meters between the first-level and second-level key's bus stop IDs. This is the same as the one in all_areas.pkl, except in that file, you can only view the bus or car distance from the bus stops in a particular area whereas bus_dists.pkl allows you to access all the distances directly.

bus_graph.pkl and car_graph.pkl

These files contain igraph.Graph objects for the bus and car graph. Refer to the pyhon igraph documentation on how to use this graph.