Content uploaded by Indaco Biazzo
Author content
All content in this area was uploaded by Indaco Biazzo on Feb 07, 2022
Content may be subject to copyright.
CityChrone: an interactive platform for
transport network analysis and planning in
urban systems
Indaco Biazzo1
Politecnico di Torino, Corso Duca degli Abruzzi 24, 10129, Torino, Italy,
indaco.biazzo@polito.it,
WWW home page: http://indacobiazzo.me
Abstract. Urban systems studies in the last decades have greatly ben-
efited from the digital revolution and the accumulation of a massive
amount of data. Extracting useful information from these data calls for
new and innovative theoretical and computational approaches. This work
presents an open-source, modular, and scalable platform for urban plan-
ning and transports network analysis, the CityChrone [citychrone.org].
The platform shows, on interactive maps, measures of performances of
public transport in cities. The measures are based on the computation
of the travel time distance between a large set of points. Thanks to the
high efficiency of the routing algorithm developed, the platform allows
users to create new public transports networks and showing the effect on
mobility in a small amount of time. A preliminary analysis of the user-
generated scenarios is presented. All the source code of the CityChrone
platform is open-source, and we employ only open data to ensure the
reproducibility of results.
Keywords: public transport networks, accessibility measures, urban
systems, urban planning, temporal networks
1 Introduction
In 2007, for the first time in human history, the population living in urban
areas exceeded the global rural population[26]. Moreover, the urbanization pro-
cess goes on, and it is expected that in 2050 two-third of the world population
will live in cities[26]. People in cities are connected thanks to a large number
of processes and different interactive networks. The city is a highly non-linear
and out of equilibrium process[19], and problems that institutions, stakehold-
ers, and private citizens will have to face will be more and more related to this
high level of complexity. Developing sustainable and efficient citizen mobility
and commuting systems is one of the most widespread challenge. Nowadays,
thanks to the ICT (Information and Communication Technologies)[40], and the
subsequent massive quantities of data accumulated gathered the attention of the
scientific community and fostered the emergence of many quantitative studies
2 Indaco Biazzo
aiming at identifying statistical patterns behind the dynamics of humans mo-
bility within or between cities [37,31, 50, 15], as well as on their infrastructures
and services [38, 27, 29, 47, 32, 48, 49]. Data about cities and their inhabitants’
habits are nowadays collected and available for research and commercial pur-
poses. The information extracted from the statistical analysis of the properties
coming from mobility-related data can significantly impact everyday life, helping
citizens perform better choices in terms of more environment-friendly mobility
solutions, more efficient movements in general, and optimal choice of the place
to live. Generally speaking, more precise and easy-to-understand information
about the criticality or efficiency of transport services in urban environments is
essential at each level of modern society (private citizens and companies, pub-
lic administrations, and research institutes). This work presents the CityChrone
project [www.citychrone.org], an interactive public transport network analysis
and planning platform. The analysis is based on accessibility measures. The sci-
entific community defined the concept of accessibility several decades ago[33, 21,
22], in order to give a precise quantification of the performance of transportation
systems per se and with other aspects of people’s lives. Despite its importance,
there is not a unique possible definition of accessibility: this could depend on the
availability of data or the aim of the researchers performing the analysis [46, 21,
41, 30]. On the other hand, such dispersion increases the difficulty for a straight-
forward interpretation of accessibility metrics, preventing their operational use
by policymakers. Recently, new metrics have been proposed [20] in a general
framework aiming to provide a unified point of view in which the temporal di-
mension is at the core. These new metrics are robust and general enough to be
applied to different urban systems and different means of transportation. Thanks
to a crucial modification to an existing very efficient routing algorithm for public
transports, the CityChrone platform can compute those quantities in a fast way
for public transports. Users can explore different scenarios of public transports
networks in nearly real-time (the time needed to re-compute all the accessibility
quantities is less than two minutes for medium-sized cities). The CityChrone
platform is an open-source project, and it is published on GitHub[1]. The aim
is to involve institutions, companies, and private citizens interested in develop-
ing an interactive platform for city knowledge, awareness, and planning. The
CityChrone project aims to facilitate the way scientific results are presented,
enhanced usability and comprehension through an interactive platform on the
web. Moreover, particular care has been devoted to using open data or data freely
downloadable to ensure the reproducibility of the results. The work is organized
in the following way: In the first section, we describe related works, and in the
second section, the routing algorithm. Then in the third section, we present the
data and preprocessing procedures. In the fourth section the citychrone platform
is presented. In the last section, before the conclusion, a preliminary analysis of
the public transport network created by users is shown.
CityChrone 3
2 Related works
The literature about accessibility measures is vast and started several decades
ago, as stated in the introduction. Despite this vast production of theoretical
tools, the computation of these quantities in a real case scenario and in more
than one particular case are very few. Moreover, even less interactive platforms
have been created to visualize these quantities. One of the first web-based plat-
forms was mapnificien [2]. In this platform, the user can choose a city and vi-
sualize isochrones, computed considering displacements with public transports,
selecting a starting point on the map. Mapnificien does not show accessibility
maps. The isochrones are computed only between stops, and the walking area
is only roughly estimated. A platform for computing and visualize accessibility
measures is described in [42]. The authors define an accessibility measure and an
interactive platform to show several informative layers. The proposed measure
is not based on actual travel time but instead on a public transport network’s
”centrality” measure. Each city stop has a score based on several factors, such as
the velocity of transportation, the distance from train stations, and the capacity.
This measure has more than five free parameters with no constraints, reducing
the measure’s universality and transparency. The platform should be accessible
for one case study, the Baltimore-Washington DC region, but the URL of the
demo is not reachable [3]. No source code is available. Another example of ac-
cessibility measure is described in [22]. The authors introduced an accessibility
measure that gives four possible values to the public transport stops, from low
to very high, based on frequency and means of transportation of public trans-
ports routes passing through the considered stop. No source-code or web-based
platform was released. Closer to our approach is the ”The Metropolitan Chicago
Accessibility Explorer” [51]. This platform shows accessibility measures based on
isochrones. In the platform[4] the user can choose several different layers show-
ing, for instance, the number of jobs or other services reachable within a given
time for each census block of the city. The travel time is computed thanks to
the OpenTripPlanner library [cite], an open-source routing library. The travel
times are precomputed, and the platform shows the results. The source code
was not released. The CityChrone platform, described in this work, has several
key differences from all the above platforms. CityChrone uses only open data
and standard processes that can be easily applied to every urban system where
public transports data are available. Moreover, as far the author knows, it is the
only available platform, that thanks to the efficiency of the routing algorithm
used, users can build new scenarios of public transports and visualize the effect
on accessibility measures after a small amount of time (less than 2 minutes for
medium-size cities like Rome and Boston).
3 Accessibility quantities
The accessibility quantities considered are presented in [20].Here we give just
a quick review. We want to measure the performance of public transports to
4 Indaco Biazzo
explore the space and connect people in a city. The starting idea is that the
isochrones, i.e., the surface at equal time distance tfrom a starting point p0at
time t0, see Fig.1, could be used to measure the performance of public transports
in cities. More extensive is the area of isochrones larger is the portion of the city
that is possible to explore given a time tfrom the starting point. Based on that,
we define two measures: the velocity score and the sociality score.
The velocity score, given a starting point p0at time t0, measure the velocity of
exploring the space around the point considered. This measure can be interpret
as the average velocity taken a random directions of displacement given a typi-
cal travel time. The precise definition is the following: consider the covered area,
A(t, (p0, t0)) of a isochrone at time tstarting from p0at time t0. We define an
effective ray ¯r(t, (p0, t0)) as: ¯r(t, (p0, t0)) = pA(t, (p0, t0))/π. Dividing it by the
time twe obtain an effective average velocity: ¯v(t, (P0, T0)) = ¯r(t, (p0, t0))/t. The
effective average velocity is defined for every point p0in the map and starting
time t0. The velocity score is obtained averaging over the journey time distribu-
tion probability f(t):
v(p0, t0) = Z∞
0
v(t, (p0, t0))f(t)dt, (1)
The journey time distribution probability is the probability distribution of travel
time on public transport in the city [35]. Observing the velocity score of the cities
on CityChrone [citychrone.org] it is clear that the center of the city has a very
high-velocity score compared to the suburbs. People living in the center of cities
are well served by public transports, having all the directions of displacement al-
lowed, usually also with good and fast public transports, like trains and subways.
Instead, in the suburbs, only directions towards the center are well served by
public transports, and all the others have poor or no public transports services.
The sociality score measure instead the possibility of meet people starting from
the point p0at time t0. It measures the number of people reachable from p0at
time t0in a typical daily working trip. The definition is similar to the 1, where
instead of the effective velocity v(t, (p0, t0)), we take the average of the amount
of people pop(t, (p0, t0)) living inside the isochrone at time tstarting from the
point p0at time t0. This measure considers, at the same time, the public trans-
ports services and the density and distribution of population in the city. For the
exact definitions, robustness as well as statistical analysis of these accessibility
measures on a large set of cities see[20].
4 Routing Algorithm
Calculating the accessibility measures require to compute travel time distances
by public transports between each point in a city. In general, if we consider
a grid of points covering a medium-sized city, with reasonable steps, e.g., less
than 500m, the number of points is of order ∝103and the number of jour-
neys to compute is of the order of 106−107for each city. The availability of
efficient routing algorithms is indeed mandatory. Moreover, we want a flexible
CityChrone 5
algorithm that allows for fast computation of all travel time distances between
points when the schedules change, meaning that data preprocessing time should
be reduced to the minimum. For road networks, one can compute the driving
directions in a millisecond or less at the continental scale, but it is not the case
for public transport networks [18]. The approaches and speedup techniques used
for road network routing algorithms fail [18], or they are not so effective on
public transport networks. In the last years, different approaches, not based on
the graph structure of the problem, have emerged in literature where the most
promising ones are the RAPTOR algorithm [23] and the CSA Algorithm [24].
Between them, the CSA algorithm seems to be the fastest for the computation
of the earliest arrival time [25]. Both algorithms are efficient and fast, with short
preprocessing time. Both algorithms have some limitations when considering
footpaths to change stops and means of transports, reducing the performance
and applicability of those algorithms in urban contexts. The algorithm we de-
vised, the ICSA, is based on the CSA algorithm, but we introduced a crucial
modification that allows us to use it in urban systems considering realistic foot-
paths between stops and means of transport. In the Supplementary Material [5],
we describe the CSA algorithm and then the ICSA.
5 Urban tessellation and data preprocessing
Cities have no unique and accepted way to define their area and border. More-
over, large urban systems are composed of several different public transport
operators, some of which span their operation well beyond the city’s limit, up to
national scale. In order to limit the area of analysis, we adopt the definition of
cities made by OECD/EU as ’functional economic units’[43]. It uses population
density to identify urban cores (city core) and travel-to-work flows to identify
the hinterlands whose labor market is highly integrated with the cores (commut-
ing zone). We consider in our analysis only the stops inside both regions and
only connections that connect them. In order to have uniform measures of the
performance of the public transports over the area, we tassellate the city area
by a hexagonal grid with distance center to center of nearby hexagons of 0.4km.
Then we retain only hexagons that have at least one stop reachable in 15 min-
utes by walk. For each hexagon in the city, we compute the resident population
inside. The data has been gathered through the Eurostat Population Grid [14]
for the European cities and the Gridded Population of the world made by the
Center for International Earth Science Information Network [34]. This spatial
population dataset divides the population into squares with a surface of 1 km2,
while our tessellation uses hexagons of smaller surfaces (∼0,1km2). Hence, we
assigned the population in each square to the hexagons overlapping them, pro-
portionally to the fraction of overlapping surfaces. The accessibility measures,
based on isochrones, use public transport schedules, streets networks and spatial
population distributions in cities. These data are nowadays easily downloadable
from the web. Public transport companies usually release their schedules in a
uniform way, using the GTFS format[6]. From this file it is possible to extract
6 Indaco Biazzo
the locations of the stops and the connections. The variation of public trans-
ports connections does not change significantly in the working days [16], [36],
so for each city, we choose a Wednesday in the period of validity of the GTFS
file that is not a holiday. We downloaded GTFS files from a repository [7], or
directly from the public transports company website. For each hexagon and stop
present in a city, we compute the time walking distance between all the stops
and hexagons reachable in 15 minutes by walk. The walking path are computed
by an OSRM backend[39] with street graphs taken from OpenStreetMap[44].
The walking speed was set to 5km/h.
(a) Home page (b) Isochrone layer
Fig. 1: a: home page, where it is possible to select the city to explore accessibility
quantity. b: Example of isochrone compute in the city of Paris
6 The CityChrone platform
CityChrone is an open-source web app, developed within the open-source meteor
framework[8]. The CityChrone platform is able, given the city’s tesselletion, the
population distribution, the stops, and the connections, to compute isochrones
and several accessibility quantities based on public transports and shows them
on interactive maps for every city. Moreover, the platform allows users to mod-
ify the connections adding new metro lines. After recomputing all accessibility
quantities, the user can check how the new scenario changes the accessibility
measures.
The user experience The platform has three principal sections: the starting page,
the visualization page and, the scenario page. On the starting page, fig.1a, the
user selects the city. The user, clicking on a city, is redirected to the visualization
page. In this section, the user can explore different layers that describe distinct
aspects of public transport performances. In the left sidebar, the user can select
the layers (velocity score, sociality score, population, isochrone and sociality
and velocity score difference). The isochrone layer shows the isochrones in the
city, fig.1b. If the user clicks on the map, an isochrone is shown starting from
the clicked point. If present, on the left sidebar, it is possible to choose a new
CityChrone 7
scenario with new metro lines added to the city and check the difference in the
isochrones and the accessibility quantities respect the default scenario. In some
cities, there is the button ”new scenario” by which the user goes to the scenario
page.
(a) Velocity score layer (b) Scenario
Fig. 2: a: Velocity score layer of Paris. b: Scenario ”rer+circle” create by the user
”mat” for the city of Rome with the layer of the improvements in the velocity
score.
In the scenario page users add new metro lines to the city. There is a lim-
ited amount of budget available to users to build new metro lines. The cost of
the metro lines is computed given a fixed price for each station and a price per
kilometer for the subway tube. The user can add new metro lines by clicking
on the ”add metro” button. The metro lines added can be dragged, deleted, or
expanded (making, for instance, bifurcation). When the constructions of new
subways are terminated, the user can click on the ”compute” button to com-
pute the new accessibility quantities. In the meanwhile, the user can insert his
name and the title for the scenario made. Then, ended the computation, the
visualizator page is loaded, with the scenario created selected. The page shows
the ”rank”, highlighting the position of the scenario created. The rank of the
scenarios is computed according to the average value of the velocity score per
person in the city.
Backend Particular care has been paid to the optimization of some aspect of
the platform, the visualization of large amount of data, the minimization of the
transfer of data between server and clients, and the scalability of the platform to
large audience. The server side of the CityChrone platform is able to, given the
data about the public transport of a city, to compute the accessibility quantities
and store them in a database. The principal components of the back-end are:
the data, the routing functions and the accessibility quantities, and the functions
related to the visualizations of the information on maps. Working with standard
JavaScript objects and HTML code could reduce the performance and the num-
ber of objects visualized on the map. Usually, there are about 103of hexagons
in a city reaching for Paris the value of 6 ∗104. The hexagons are too many to
be displayed in a standard browser, so we merge contiguous hexagons with the
8 Indaco Biazzo
same color. Empirically we found that their number is reduced by a factor of ten.
The most computational demanding operation is the calculation of accessibility
quantities. When a new scenario is created, the computation is made client-side
to not overload the server-side. All the routing functions are written to be used
both by client and by server-side. On the client-side, the computation exploits
the parallel computing that modern browsers allow through Web Workers [9].
The client side computation allows the CityChrone platforms to scale to many
users without the need for ample server resources. The current version of City-
Chrone runs on a virtual server with eight dedicated X86 64bit cores with 16GB
of RAM. The client-side computation of the new scenario required: i) the com-
putation of new connections added to the existing ones, given the list of new
metro lines created by users. We assume a fast metro line (see for example [10]),
with acceleration of 1.3 m/s2, a maximum velocity of 30 m/s and frequency
every 2 minutes. These parameters can easily changed to reflect local project
constraints. ii) For each new stop, the walking times with stops and hexagons
reachable in 15 minutes by walk is computed. A OSRM server [39,11] with the
street network taken from OpenStreetMap [44] of the city is active on the server.
To save server-side resources, each new stop can take the walking paths equal
to those of the nearest hexagon or stop (this could reduce the precision of the
calculation, but it avoids having an OSRM server always up). We implement
both solutions. iii) The computation of the new travel times between all points
and the relative accessibility quantities. The running time for the complete com-
putation of a new scenario is about 1-2 minutes for a medium-size city (Rome,
Boston, etc.) on a computer with a 4-core CPU. For a larger city, the time can
reach 30 minutes, and for these cities, we do not allow, so far, users to create a
new scenario.
7 Preliminary analysis of new public transport network
scenarios: the case of Rome
In this section, we analysed the scenarios created for the city of Rome. In two
events, in 2017 and 2018, the platform was presented inside two public scientific
events organized in Rome [12]. In the period 2017-2021, the web page of Rome
has had about 3000 unique views, and users created more than 350 new public
transport scenarios. The users have a budget of 5000M€of euros to construct
new metro lines. We assume that a single stops cost 100M€(see for instance
the average costs of a Paris metro station [13]) and the tube 30M€per km
(average cost of a tunnel boring machine tube [17]). These costs are assumed
valid on average for a medium size city, but locally can vary depending on several
different factor, ranging from architectural projects to soil composition. More
specific cost functions can be easily integrated in the platform. The goal is to
maximize the gains in velocity scores or sociality scores of the whole city. In
order to analyze the user solutions so far created, we cluster them according
to the impact on the accessibility measures considered, the sociality score. We
associate to each scenario an array containing the sociality score computed for
CityChrone 9
Fig. 3: The figure shows the scenarios of new subways generated by users in the
case of Rome. The solutions are clustered according to the distribution on the
city of the sociality score. The average value of the sociality score of the scenarios
in each cluster is shown in the bottom plot. The clusters are ordered according
to the best solution they contain. The four plots above show up to the six best
scenarios for the four best clusters. For each cluster, the improvements in the
sociality score of the best scenario are shown in the right plots.
each point of the hexagonal grid covering Rome, and we cluster the scenarios
according to them. We use the affinity propagation algorithm [28] implemented
in scikit-learn[45] because it does not need to specify the number of clusters
beforehand. We rank each cluster of scenarios according to the best scenario it
contains. Figure 3 shows the best four clusters, together with the six best metro
lines proposed by users (if present). All the four clusters of solutions share a
future, the need for a circular line in Rome. The best solution so far seems to
merge a circular line with high-speed connections that irradiate from Rome’s
city center. Cluster 1 shows solutions with a circular line and, usually, some
small metro lines enter the city’s center. The scenarios belonging to cluster 2
have a more complex shape, with lots of metro stations serving the city center
10 Indaco Biazzo
and partial circular lines. Cluster 3 has only one scenario, which is a circular line
with very dense stops. The solution of Cluster 4 has a circular line with a fast
connection to a west part of Rome that overlooks the sea. Closer inspections of
the scenarios are possible on citychrone.org.
8 Conclusion
In this work, we presented CityChrone: an interactive platform for public trans-
port analysis and planning. The platform shows several accessibility measures
about the performance of public transports in more than 30 cities around the
world. The accessible measures shown are based on the computations of a vast
number of travel time distances between a grid of points in the city. The plat-
form exploits a very efficient routing algorithm that allows a fast computation
of the accessible quantities. Users can explore new public transports scenarios,
build new metro lines, and check quickly (in minutes) how the proposed solu-
tions change the accessibility measures. The primary computational requests are
designed to run client-side, allowing the CityChrone platform to scale to a large
number of users with small server-side resources. New cities can be easily added
to the platform having just the schedules of the public transports. A script pro-
cess the data, and then, after loaded in the CityChron platform, it is possible to
visualize isochrones, accessibility quantities and test new transports scenarios.
The CityChrone platform is the first step of an open-source project aiming to
create a community of companies, public institutions, stakeholders, developers,
and private citizens interested in developing interactive platforms to analyze the
transports in cities and search for innovative solutions to mobility problems in
cities.
9 Acknowledge
This work has been supported by the SmartData@PoliTO center on Big Data
and Data Science.
References
1. https://github.com/CityChrone
2. https://www.mapnificent.net/
3. http://www.geosimulation.org/transit.html
4. http://urbanaccessibility.com/
5. https://bit.ly/3hkG613
6. https://developers.google.com/transit/gtfs
7. www.transitfeeds.com
8. https://www.meteor.com
9. https://html.spec.whatwg.org/multipage/workers.html
10. https://tinyurl.com/yycnsau6
11. https://github.com/Project-OSRM/osrm-backend
CityChrone 11
12. http://www.kreyon.net/kreyonDays/
13. https://marroninstitute.nyu.edu/events/how-much-does-it-cost-to- build-a-subway
14. Eurostat population grid, http://ec.europa.eu/eurostat/web/gisco/geodata/
reference-data/population-distribution-demography/geostat
15. Alessandretti, L., Aslak, U., Lehmann, S.: The scales of human mobility. Nature
587, 402–407 (11 2020)
16. Alessandretti, L., Karsai, M., Gauvin, L.: User-based representation of time-
resolved multimodal public transportation networks. Royal Society Open Science
3(7) (2016), http://rsos.royalsocietypublishing.org/content/3/7/160156
17. Attin`a, M., Basilico, A., Botta, M., Brancatello, I., Gargani, F., Gori, V., Wil-
helm, F., Menting, M., et al.: Annex 13 – case study on tunnels of ”assess-
ment of unit costs (standard prices) of rail projects (capital expenditure)” (2018),
https://tinyurl.com/48chu6r8
18. Bast, H., Delling, D., Goldberg, A., M¨uller-Hannemann, M., Pajor, T., Sanders,
P., Wagner, D., Werneck, R.F.: Route planning in transportation networks. In:
Algorithm engineering, pp. 19–80. Springer (2016)
19. Batty, M.: The size, scale, and shape of cities. Science 319(5864), 769–771 (2008)
20. Biazzo, I., Monechi, B., Loreto, V.: General scores for accessibility and inequality
measures in urban areas. Royal Society Open Science 6(8), 190979 (2019)
21. Black, J., Conroy, M.: Accessibility measures and the social evaluation of urban
structure. Environment and Planning A 9(9), 1013–1031 (1977)
22. Bok, J., Kwon, Y.: Comparable measures of accessibility to public transport using
the general transit feed specification. Sustainability 8(3), 224 (2016)
23. Delling, D., Pajor, T., Werneck, R.F.: Round-based public transit routing. Trans-
portation Science 49(3), 591–604 (2014)
24. Dibbelt, J., Pajor, T., Strasser, B., Wagner, D.: Intriguingly simple and fast tran-
sit routing. In: International Symposium on Experimental Algorithms. pp. 43–54.
Springer (2013)
25. Dibbelt, J., Pajor, T., Strasser, B., Wagner, D.: Connection scan algorithm. CoRR
abs/1703.05997 (2017), http://arxiv.org/abs/1703.05997
26. of Economic, D., Social Affairs, P.D..: World urbanization prospects: The 2014
revision, highlights (st/esa/ser.a/352). Population Division, United Nations (2014)
27. Fleurquin, P., Ramasco, J.J., Egu´ıluz, V.M.: Characterization of delay propagation
in the US air-transportation network. Transportation journal 53(3), 330–344 (2014)
28. Frey, B.J., Dueck, D.: Clustering by passing messages between data points. science
315(5814), 972–976 (2007)
29. Gallotti, R., Bazzani, A., Rambaldi, S.: Understanding the variability of daily
travel-time expenditures using GPS trajectory data. EPJ Data Science 4(1), 18
(2015)
30. Geurs, K.T., Van Wee, B.: Accessibility evaluation of land-use and transport strate-
gies: review and research directions. Journal of Transport geography 12(2), 127–140
(2004)
31. Gonzalez, M.C., Hidalgo, C.A., Barabasi, A.L.: Understanding individual human
mobility patterns. nature 453(7196), 779–782 (2008)
32. Guimera, R., Mossa, S., Turtschi, A., Amaral, L.N.: The worldwide air transporta-
tion network: Anomalous centrality, community structure, and cities’ global roles.
Proceedings of the National Academy of Sciences 102(22), 7794–7799 (2005)
33. Hansen, W.G.: How accessibility shapes land use. Journal of the American Institute
of planners 25(2), 73–76 (1959)
12 Indaco Biazzo
34. for International Earth Science Information Network CIESIN Columbia Univer-
sity, C.: Gridded Population of the World, Version 4 (GPWv4): Population Count
(2016), http://dx.doi.org/10.7927/H4X63JVC
35. K¨olbl, R., Helbing, D.: Energy laws in human travel behaviour. New Journal of
Physics 5(1), 48 (2003)
36. Kujala, R., Weckstr¨om, C., Darst, R.K., Mladenovi´c, M.N., Saram¨aki, J.: A collec-
tion of public transport network data sets for 25 cities. Scientific Data 5, 180089
(2018)
37. K¨olbl, R., Helbing, D.: Energy laws in human travel behaviour. New Journal of
Physics 5(1), 48 (2003)
38. Li, R., Dong, L., Zhang, J., Wang, X., Wang, W.X., Di, Z., Stanley, H.E.: Simple
spatial scaling rules behind complex cities. Nature communications 8(1), 1–7 (2017)
39. Luxen, D., Vetter, C.: Real-time routing with openstreetmap data. In: Proceed-
ings of the 19th ACM SIGSPATIAL International Conference on Advances in Ge-
ographic Information Systems. pp. 513–516. GIS ’11, ACM, New York, NY, USA
(2011), http://doi.acm.org/10.1145/2093973.2094062
40. Mayer-Sch¨onberger, V., Cukier, K.: Big data: A revolution that will transform how
we live, work, and think. Houghton Mifflin Harcourt (2013)
41. Miller, H.J.: Measuring space-time accessibility benefits within transportation net-
works: basic theory and computational procedures. Geographical analysis 31(1),
1–26 (1999)
42. Mishra, S., Welch, T.F., Torrens, P.M., Fu, C., Zhu, H., Knaap, E.: A tool for
measuring and visualizing connectivity of transit stop, route and transfer center in
a multimodal transportation network. Public Transport 7(1), 77–99 (mar 2014)
43. OECD: Redefining ”Urban”. OECD Publishing (2012), https://www.oecd-ilibrary.
org/content/publication/9789264174108-en
44. OpenStreetMap contributors: Planet dump retrieved from https://planet.osm.org
. https://www.openstreetmap.org (2017)
45. Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O.,
Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V., Vanderplas, J., Passos, A.,
Cournapeau, D., Brucher, M., Perrot, M., Duchesnay, E.: Scikit-learn: Machine
learning in Python. Journal of Machine Learning Research 12, 2825–2830 (2011)
46. Schm¨ocker, J.D., Quddus, M.A., Noland, R.B., Bell, M.G.: Mode choice of older
and disabled people: a case study of shopping trips in London. Journal of Transport
Geography 16(4), 257–267 (2008)
47. Sen, P., Dasgupta, S., Chatterjee, A., Sreeram, P., Mukherjee, G., Manna, S.:
Small-world properties of the Indian railway network. Physical Review E 67(3),
036106 (2003)
48. Sun, X., Wandelt, S., Hansen, M.: Airport road access at planet scale using pop-
ulation grid and openstreetmap. Networks and Spatial Economics 20(1), 273–299
(2020)
49. Sun, X., Wandelt, S., Zhang, A.: Comparative accessibility of chinese airports and
high-speed railway stations: A high-resolution, yet scalable framework based on
open data. Journal of Air Transport Management 92, 102014 (2021)
50. Yan, X.Y., Zhao, C., Fan, Y., Di, Z., Wang, W.X.: Universal predictability of mo-
bility patterns in cities. Journal of The Royal Society Interface 11(100), 20140834
(2014)
51. Yin, S., Li, M., Tilahun, N., Forbes, A., Johnson, A.: Understanding transportation
accessibility of metropolitan chicago through interactive visualization. In: Proceed-
ings of the 1st International ACM SIGSPATIAL Workshop on Smart Cities and
Urban Analytics. pp. 77–84. ACM (2015)
