Content uploaded by Geoff Boeing
Author content
All content in this area was uploaded by Geoff Boeing on Jan 15, 2021
Content may be subject to copyright.
The Right Tools for the Job:
The Case for Spatial Science Tool-Building
Geoff Boeing
Department of Urban Planning and Spatial Analysis
Sol Price School of Public Policy
University of Southern California
August 2020
Abstract
This paper was presented as the 8
th
annual Transactions in GIS plenary address at the
American Association of Geographers annual meeting in Washington, DC. The spatial
sciences have recently seen growing calls for more accessible software and tools that
better embody geographic science and theory. Urban spatial network science offers one
clear opportunity: from multiple perspectives, tools to model and analyze nonplanar
urban spatial networks have traditionally been inaccessible, atheoretical, or otherwise
limiting. This paper reflects on this state of the field. Then it discusses the motivation,
experience, and outcomes of developing OSMnx, a tool intended to help address this.
Next it reviews this tool’s use in the recent multidisciplinary spatial network science
literature to highlight upstream and downstream benefits of open-source software
development. Tool-building is an essential but poorly incentivized component of aca-
demic geography and social science more broadly. To conduct better science, we need to
build better tools. The paper concludes with paths forward, emphasizing open-source
software and reusable computational data science beyond mere reproducibility and
replicability.1
1. The Wrong Tools for the Job
Do I need to know the precise polygonal geometries of Los Angeles and the University of
Southern California to assert that the latter is within the former? No. My mind contains no
such precise geometric model of points and lines, yet I know that USC is in Los Angeles. When
humans reason with the real world, they focus on its objects, relations, and processes—rather
than starting with geometry—because these are the keys to understanding and explaining
1
Preprint of: Boeing, G. 2020. The Right Tools for the Job: The Case for Spatial Science Tool-Building.
Transactions in GIS, published online ahead of print. doi:10.1111/tgis.12678. Email: boeing@usc.edu
1
the real world. Our GIS tools, however, usually do the opposite. Built from the geometry-up
around the legacy logic of traditional cartography (geometries and layers), most GIS tools
today are restricted by that legacy’s limited ability to model objects, relations, and processes.
A representational tension thus exists in GIScience between being a geometric information
science versus an ontological,relational, and processual information science.
Computational tools help us reason with the world outside. Accordingly, their represen-
tations of reality should start with domain theory—well-substantiated systems of ideas to
understand and explain phenomena—rather than the constraints of a certain technology or
computing platform (Gahegan, 1999, 2018). For geographic research questions, the relevant
domain theory often utilizes object-oriented relations and processes, rather than Cartesian
abstractions of space and geometry, even if our computational tools cannot. We need tools
that fundamentally embody appropriate scientific theory rather than twisting theory to fit
within their representational and computational limitations (Harris et al., 2017; Poorthuis
& Zook, 2019). Although new data sources and knowledge discovery systems can help us
wrestle with tricky questions, we impoverish our ability to reason with computers if we do
not center theory when we create computational representations of the real world—even if
we must rethink or advance our technologies and tools to do so.
It is relatively easy to level such critiques, but if we want better GIS tools to study so-
ciospatial objects, relations, or processes, we need to build them. Urgent tool-building op-
portunities exist today across many geographic subdomains. As but one interdisciplinary
example, consider recent advances in spatially-informed graph models. Modeling spatial
dynamics, relations, and topology too often took a backseat historically to geometry, but
graphs offer possible ways forward. For instance, geographic knowledge graphs allow us
to build spatial information systems around objects and relations
1
rather than geometries,
to better answer ontological spatial questions (Yan et al., 2019). Yet such tools remain in
their infancy today. As another example—and the motivating example on which this paper
focuses—graph models of city transportation networks allow topological and dynamical
inquiry into urban processes, flows, and structure (Barthelemy, 2011; Marshall et al., 2018).
Yet such tools traditionally relied on network geometry rather than topology (due to data
availability and computational constraints), incorporated domain theory poorly, and were
usually ad hoc rather than generalizable, accessible, and reusable (Boeing, 2017, 2018).
To conduct better science, we need to build better tools. Such tool-building allows aca-
demics to better operationalize and hypothesis-test theory and therefore forms an essential—
but poorly incentivized—pillar of scholarly research. In this paper, I reflect on my own
tool-building experiences in urban planning and geography: facing the need for a better
tool to model and analyze urban street networks in a scalable, theoretically-sound way, I
developed a new open-source Python-based software package called OSMnx. This paper
considers its history, motivation, and purpose, then reviews its recent use in the empirical
street network science literature. In turn, this illustrates the utility of academic tool-building
and its downstream—and upstream—value. This paper concludes by proposing better align-
2
ment of academic incentives with the positive externalities of conducting open science and
developing open-source spatial research software.
2. If You Want Something Done Right...
Urban science, sitting at the intersection of city planning, geography, and computational data
science, aims to advance our knowledge of cities’ fundamental patterns and relationships
by modeling spatial big data (Batty, 2013; Mattern, 2013; Solecki et al., 2013; Kitchin, 2016;
Sallis et al., 2016; Alberti, 2017; Acuto et al., 2018; Kontokosta, 2018; Barthelemy, 2019; Batty,
2019; W. Kang et al., 2019; Lobo et al., 2020). Despite urban science’s recent bold claims to
a “new kind of science,” urban geographers, sociologists, and planners have of course long
investigated cities’ patterns and processes through spatial data, mathematical models, and
the scientific method (Burgess, 1925; Hoyt, 1951; Branch, 1966; Batty, 1971, 1980; Lee, 1973,
1994; Bertuglia et al., 1998; Johannesen et al., 1998; O’Sullivan & Manson, 2015; Behrend &
Levin-Keitel, 2019; Derudder & van Meeteren, 2019). Computational geography itself now
has a long history, yet, too often, geographic science and domain theory fail to fully permeate
our computational tools (Gahegan, 1999; Harris et al., 2017; Arribas-Bel & Reades, 2018;
Gahegan, 2018; Singleton & Arribas-Bel, 2019; Gahegan, 2020).
Why is that? Reflecting on these insufficient links between GIScience—the scholarly
field—and GISystems—the software and tools, Gahegan (2018) highlights two themes of
particular relevance here. First, he argues that the GIScience research community usually
does not develop its own software tools because it is in nobody’s short-term interests to do
so. There are of course exceptions to this rule (some of which will be discussed later) but
in general too little geographic science and theory make their way into reusable, accessible
tools due to misaligned incentives, expectations, and training in academia. Second, and in
turn, Gahegan argues that GIScientists must foster a more robust software development
community to build and democratize better scientific research tools that are accessible and
available to everyone.
Geography journals have witnessed a recent surge of attention to this under-appreciated
importance of academic tool-building. For instance, Poorthuis & Zook (2019, p. 8) argue
that “as a discipline we need to take charge of building and maintaining our own software
platforms. These platforms should be open, accessible, and modifiable by the entire academic
community and reflect the diversity and heterogeneity of our discipline.” Yet, to date, such
tools rarely materialize in practice because, as Gahegan (2018, p. 24) puts it, we have an
“academic culture that fails to reward those who build or maintain tools and software and
encourages a short-sighted and individualistic approach to research.” Along similar lines, Rey
(2019, p. 7) recalls as a junior scholar being told to stop developing tools because “‘You need
to be writing papers.’” He continues, “My colleagues were being brutally honest and trying to
reign in my idealism so that my efforts were more aligned with the realities of promotion and
tenure cases at the time.”
3
The widespread disincentivization of academic tool-building produces several negative
outcomes. First, most of our tools rely on impoverished representations of geographic
theory because our theoreticians have little incentive or training to build tools. Second, most
scientific computational workflows exist only as ad hoc scripts to answer a specific research
question before being shelved, rather than being generalizable, documented, shared, and
accessible.
2
An enormous amount of scholarly effort is wasted as we endlessly reinvent each
others’ wheels. Third and accordingly, reproducibility and replicability remain an outstanding
challenge (Brunsdon, 2016; Kedron et al., 2019; Koster & Rowe, 2020). This has become a
key motivation for the open-source and open-science movements (Rey, 2009; Donoho, 2017).
But as a scientific community we need to go beyond mere reproducibility and replicability to
consider the public reusability of our tools and workflows. Otherwise we fail to unlock the
broader benefits and spillover effects of tool-building.
These problems plague most geographic disciplines, including the study of cities. Today,
a consensus is growing around the importance of harnessing GIScience to the open-source
and open-science movements to make urban scientific research more tractable, replicable,
theoretical, impactful, and approachable for non-computer scientists. Yet urban science too
often lacks open data sources and reusable, accessible, theoretically-sound tools. Incentivizing
authors to share their data and computational workflows when submitting journal articles
is one nascent step in the right direction. Incentivizing and building reusable, accessible,
theoretically-sound tools is another.
3. Tools for Street Network Science
Many of these challenges hit close to home for me. A few years ago, I wanted to conduct a
nationwide study of US street network form to better understand the fundamental charac-
teristics and outcomes of different urban planning paradigms. How did network structure
change as planners reoriented cities around the spatial logic of the automobile? And what do
we see in recently-built neighborhoods, considering the rise of neotraditional urban design
practice? I was interested in the geometry of these networks, but more important was their
topology and the sociospatial dynamics they underlie and organize.
Initially I assumed that some tool must already exist to automatically construct nonplanar
directed graph models from ubiquitously available street network topology data. I was wrong
on both counts. On the tool side, while some scholars had previously studied similar topics,
no one’s research software appeared to be publicly available, well-documented for reuse, or
sufficiently scalable. Meanwhile, on the data side, mapping platforms like Google Maps did
not offer their spatial network data for download. What was ubiquitously available was the
US Census Bureau’s TIGER/Line roads shapefiles, but they place primacy on geometry and
contain insufficient topological information to properly model nonplanar networks (Boeing,
2018). I eventually turned to OpenStreetMap.
First launched in 2004, OpenStreetMap is a wiki-style worldwide mapping project and
geospatial data repository with good coverage and quality (Girres & Touya, 2010; Haklay,
4
2010; Corcoran et al., 2013; Zielstra et al., 2013; Barron et al., 2014; Maier, 2014; Basiri et
al., 2016; Sehra et al., 2019). To date, over 1 million different users have contributed content
including 6 billion nodes (i.e., geospatial points), 600 million ways (i.e., geospatial lines and
boundaries), and related descriptive data. It is not a perfect data source: researchers estimate
that >95% of OpenStreetMap contributors are male, suggesting the possibility of correlated
biases in content creation (Schmidt & Klettner, 2013; Graham et al., 2015). Nevertheless,
OpenStreetMap is public, free, and an Open Source Initiative affiliate. Volunteers provide
some editorial oversight of edits, but anyone may edit the map using tools such as ESRI’s
ArcGIS Editor for OpenStreetMap. OpenStreetMap contains data on streets and highways,
transit systems, building footprints, parks and plazas, pedestrian and bicycle infrastructure,
political boundaries, and more (though non-road coverage varies around the world). But
best of all, unlike a shapefile, OpenStreetMap’s data model centers spatial objects and their
relations, including both geometric and nonplanar topological information.
Researchers typically access OpenStreetMap data through its Overpass API or by down-
loading a prepackaged regional extract from third-party organizations like Geofabrik. These
offer easily ingested street data, but the spatial networks’ topological relationships require
substantial processing to generate a useful graph model. Researchers often have to write
hundreds or thousands of lines of ad hoc code to process the data into a graph and conduct
algorithmic analyses for a one-off study. Dozens of small computational and modeling de-
cisions inevitably go unreported in the subsequent peer-reviewed literature, yet these can
drastically impact interpretation and replication when every research team codes its own
models and analytics. Would it not be better to collectively contribute to and share a reusable,
accessible, theoretically-sound set of scientific tools?
One obstacle limiting this tool landscape is that high barriers to entry exist for all but
those fluent in computer science and domain theory. When I first reviewed the urban spatial
networks literature, I was struck by how many modeling methodologies made unexplained
or even unjustified assumptions about either urban theory or spatial network theory. The
landscape lacked tools to handle nonplanar representations of space, which most spatial
networks require due to overpasses and underpasses. For simplicity, many studies resorted
to undirected graph models, which work fine for studies of form but poorly for studies of
flows that obey directionality constraints. The precise handling of common street network
features (such as self-loops, parallel edges, or culs-de-sac) often went undocumented. More
fundamentally, the Overpass API was cumbersome to work with directly to pipe data into an
appropriate model for spatial/network analysis.
In this context I began to develop what eventually became OSMnx. The tool itself is
documented
3
in detail elsewhere (Boeing, 2017), and tutorials and usage examples
4
are avail-
able online, but I briefly summarize its functionality here. OSMnx is a Python package for
collecting spatial network data from OpenStreetMap then automatically constructing a di-
rected nonplanar graph model. It is built on top of the open-source geospatial Python stack
(which I will return to later). OSMnx’s key feature is its ease of use. With just one line of code,
the researcher can download and model the street network of any study site in the world:
5
cities, towns, neighborhoods, boroughs, counties, states, nations—any spatial boundary that
OpenStreetMap has in its database. The researcher can specify the site’s drivable, walkable, or
bikeable network. OSMnx includes a suite of visualization tools and graph-theoretic analytics
(both geometric and topological) for common transportation planning, urban design, and
network science research questions. It can also automatically collect and model elevation,
building footprint, and points-of-interest data. The code is documented and open-source, so
its formal representations of theory are not black-boxes.
This project began as a few lines of Python code in a Jupyter notebook, before being
collected into a module, and later refactored into a formal package distributed online. I am
not a software engineer per se, but I knew how to code and was conversant in the relevant
domain theory. But, in the midst of working on this project for months on end, I realized that
the next person interested in similar empirical questions would face the exact same laborious
tool-building process I was then struggling through. Accordingly, I found it useful to make
OSMnx open-source for three primary reasons. First, it makes empirical work easier to review
and reproduce. Second, it allows anyone else to contribute to the tool’s ongoing development.
As discussed later, other researchers desiring useful extensions to its functionality have been
willing to add them to the codebase. Third, it empowers others working in urban science and
planning to advance their empirical research on real-world spatial networks with a reusable,
accessible, theoretically-sound tool. I discuss these latter outcomes in the following section.
4. Empirical Street Network Science with OSMnx
The teleology of tool-building suggests that the real value lies in the end use of the tool, rather
than in its origins. The purpose of developing OSMnx was to conduct empirical research on
urban form, travel dynamics, and the topological structure of transportation infrastructure.
Since its public release three years ago, several such studies have been conducted—by myself
and many others—using OSMnx for model generation, indicator calculation and visualization,
and trip simulation. To illustrate the downstream benefits of tool-building, this section briefly
reviews the recent empirical literature that uses OSMnx.
Across a variety of academic disciplines and study sites, researchers have recently used
OSMnx to download data, generate models, and analyze real-world transportation networks.
For example, Hofer et al. (2018a,b) model the street network of Graz, Austria to simulate
CO
2
emissions and traffic congestion and avoidance behavior using mobility data. Saha et
al. (2019) model Mesa, Arizona’s streets to generate a synthetic feeder network for electrical
distribution. S. Wang, Gao, et al. (2018) model Washington DC’s street network to develop
a geoprocessing framework for optimizing the meetup locations of multiple people under
congested traffic conditions. Liu et al. (2020) model Beijing’s walkable street network to
explore spatial patterns of residents’ daily leisure activities. Natera Orozco et al. (2019) model
“as-is” bicycle networks to demonstrate how cities can make small but targeted infrastructure
investments to significantly increase their connectivity and directness. Dumedah & Eshun
(2020) model Ghanaian street networks to investigate paratransit service coverage through
6
GPS data. Riascos & Mateos (2020) model Manhattan’s street network for their study of more
than one billion taxi trips in New York.
4.1. Investigating New Urban Technologies
Studies such as these often investigate the frontier of new transportation and smart cities
technologies, including autonomous vehicles, electric vehicles, ride-sharing, and bike-sharing.
Beirigo et al. (2018) use OSMnx to model service levels, operational and infrastructure
costs, and fleet utilization in hybrid street networks with both autonomous-ready and not
autonomous-ready zones. Q. Lin et al. (2018) model Manhattan’s street network alongside
travel demand data to optimize ride-share routing. Luo et al. (2020) model Shanghai’s street
network to predict demand for electric vehicle sharing systems, while Zhang et al. (2019)
model Shanghai’s bicycle network to propose a framework for planning dockless bike-sharing
services’ geofences.
4.2. Network Structure and Urban Centrality
Other studies look for fundamental relationships between topological structure—particularly
network centrality and robustness—and travel patterns and land use. M. Wang et al. (2020)
use OSMnx to model Atlanta’s street network to investigate ride-sharing accessibility as a
function of network centrality and structure. S. Wang, Xu, & Guo (2018) model Shenzhen’s
street network to explore the relationship between street centrality and land use intensity.
Hellervik et al. (2019) develop a preferential centrality measure to predict urban activity based
on street network structure. D’Angelo et al. (2017) model the street network of Fano, Italy
to identify locations of high betweenness centrality. Masias et al. (2019) operationalize a
spatial capture–recapture methodology to model social media users as a function of walkable
street network centrality indicators. Dingil et al. (2019) compute and visualize indicators
of connectivity, centrality, and clustering across 86 urban areas worldwide to identify the
role of network design in easing traffic congestion. Torres et al. (2019) and Morelli & Cunha
(2019) use centrality indicators to measure street networks’ vulnerability to perturbation in
Mexican and Brazilian cities respectively (see also Baumann & Keupp, 2020; Sohouenou et al.,
2020).
4.3. Computer Science Methodological Research
Computer scientists and statistical physicists often adopt urban transportation networks as
tractable, real-world systems that can be well-represented by graphs. OSMnx has been used
accordingly to generate input graphs and feature sets for methodological research in machine
learning and network algorithms (Yin et al., 2020; Young & Eccles, 2020). Ren et al. (2019)
model Chengdu’s street network then predict traffic flow with a deep spatiotemporal residual
neural network. Law & Neira (2019) model the street networks at the centers of 100,000
cities worldwide to train a convolutional autoencoder to analyze network structure. O’Keeffe
et al. (2019) train a recurrent neural network to reproduce vehicular mobility patterns, using
7
taxi data and street network models. Martínez Mori & Samaranayake (2019) model several
cities’ street networks to empirically demonstrate heuristic approximation algorithms that
make certain network analyses computationally tractable. Feng & Porter (2020) model cities
around the world to explore their topology through persistent homology. Samson et al. (2018)
model Filipino cities to develop a genetic algorithm for optimizing paratransit services in
developing countries. Senturk & Kebe (2019) model Turkish cities to develop a heuristic
solution to a clustered variant of the classic traveling salesman problem. Neukart et al. (2017)
model Beijing’s street network to develop a quantum annealer for traffic flow optimization
on hybrid quantum computing hardware.
4.4. Indicator Calculation and Visualization
Other researchers have used OSMnx for automated indicator calculation and visualization.
Brandily & Rauch (2018) calculate street network indicators in 1,800 towns across sub-
Saharan Africa to explore the relationship between street density and population growth.
Holub (2017) calculates and visualizes indicators of bicycle network structure and connec-
tivity in Austin, Charlotte, Columbus, and Minneapolis. Quistberg et al. (2019) calculate
transportation network indicators to build a new platform for conducting cross-country ur-
ban health studies. Van Etten et al. (2019) and Verendel & Yeh (2019) visualize street network
characteristics across their respective studies’ sites. Boeing (2020a) models the street networks
of every US urbanized area, city/town, and Zillow-defined neighborhood to calculate dozens
of indicators across tens of thousands of study sites at multiple spatial scales, then shares
these models and indicators in a public repository. S. Wang et al. (2020) calculate network
structure indicators to compare American and Chinese cities. Natera Orozco et al. (2020)
compute quality-of-life indicators in Budapest by modeling its pedestrian network and local
amenities.
4.5. Urban Morphology
Several morphological studies of urban form, sprawl, and density have used OSMnx as
well. Gervasoni et al. (2017) measure urban sprawl and network-constrained destination
accessibility. Bristow (2019) compares alternative building density metrics in the context of
urban morphology. Abdelkader et al. (2018) and Boeing (2018) measure urban nonplanarity
to retheorize how planar assumptions impact street network analyses and to suggest better
models. The World Bank collects building footprint data then trains a random forest model
to identify OpenStreetMap coverage gaps that can inform future crowd-sourcing efforts and
mapping campaigns (Jones, 2019). Boeing (2020b) explores the growing role of big data in
computational urban morphology and visual analytics (Figure 1). Boeing (2019b) conducts a
cluster analysis of urban street networks around the world to theorize how planners produce
different forms of geometric spatial order, illustrated in Figure 2. Coutrot et al. (2020) build on
this theory to explore how spatial order impacts a city’s residents’ spatial navigation, finding
that residents of more grid-like places demonstrate worse ability.
8
Figure 1.
One square mile of different cities’ street networks, held at the same scale to compare
the urban form and grain (Boeing, 2020b).
4.6. Network-Constrained Trip Simulation
Finally, several studies have used OSMnx for simulating trips along a network. Waddell et
al. (2018) build regional planning models that integrate land use simulation, travel demand
modeling, and traffic assignment. Hernández-Hernández et al. (2019) calculate commute
routes and trip distances in a study of motorists’ emotions and expressions of anger in Mexico
City. Boeing (2019a) compares driving versus walking route circuity by simulating millions
of trips across 40 city street networks. Merchán et al. (2020) and Merchán & Winkenbach
(2019) investigate last-mile logistics circuity in São Paulo and develop better approximation
algorithms for urban route distance prediction. Padgham et al. (2019) examine hospital siting
by simulating network-constrained stroke service center catchment basins, while D. Lin et
al. (2019) identify the determinants of bicycle catchment basins around Shanghai’s metro
stations. Liao et al. (2020) model street networks in several world cities to compare travel time
by personal automobile versus public transit. Finally, several recent studies have used OSMnx
during the global COVID-19 pandemic for spatial epidemiology and healthcare accessibility
modeling (e.g., Adler et al., 2020; J.-Y. Kang et al., 2020).
9
4.7. Summary
The preceding survey of recent studies covers a wide range of spatial topics and disciplines,
from urban design to public health to transport engineering to computer science. These
research projects investigate various travel modes, from walking to cycling to driving to ride-
hailing. They study neighborhoods and cities in every inhabited continent. Most importantly,
I could not have conducted most of these studies myself as they exceed my knowledge and
skill, but by making OSMnx open-source and accessible, it has percolated into others’ research
designs around the world. Academic tool-building thus entails upstream and downstream
benefits for our field’s wider endeavor of scientific discovery and theory building.
5. Tool-Building in Academia
A couple of years ago, I gave a talk which partly dove into OSMnx development and the
benefits of academic tool-building and open science. At dinner that evening a senior faculty
member asked if my development work continued. I said yes and listed a few features in
development that would unlock exciting new spatial network analytics and research questions.
The faculty member pushed back on this and—echoing Rey (2019)’s recollections as a junior
scholar—suggested that if I were to continue in academia, that I must give up tool-building. To
paraphrase: “You will become known merely as a tool builder rather than a serious scholar. A
serious scholar cannot waste time on anything but empirical research and advancing theory.”
Must scholars eschew tool-building in order to successfully further their field? I believe,
rather, that we impoverish our field if we do not make our otherwise ad hoc, one-off research
tools “open, accessible, and modifiable,” to return to Poorthuis and Zook’s earlier words. In
particular, there are three primary benefits of open-source tool-building for academics: 1)
unlocking your own empirical research, 2) advancing the collective scientific and theoretical
endeavor, and 3) impacting a broader audience.
5.1. Unlocking Individual Research
First, tool-building unlocks your own research. Nearly all of us in the spatial sciences do
some tool-building as we create simple macros, scripts, or libraries of code for routinizing
mundane data processing tasks or fitting models or generating graphical and tabular output
for subsequent publication. In this era of big data and ubiquitous computation, it is inefficient
and limiting to rely on point-and-click interfaces to conduct scientific research. Coding
skills thus increasingly appear in spatial science curricula as necessary components of any
GIS skillset. Fostering such abilities is crucial as data science and coding grow central to
spatial analysis. But most importantly, as researchers motivated to ask new questions and
develop new methods, we cannot assume that theory-rich tools already exist to answer today’s
important questions. We must build the right tools to answer the right questions—and doing
so can open up new personal research trajectories.
11
5.2. Advancing the Discipline
Second, once we have built these tools, they should be open-source and accessible to advance
the wider discipline. Fortunately this philosophy already has traction in the GIScience and
urban science communities. Efforts are underway today to build a national geospatial software
center (Geospatial Software Institute, 2020) and other examples abound. Luc Anselin’s GeoDa
and GeoDaSpace software for spatial analysis and econometric modeling are free and open-
source (Anselin et al., 2006), as is Serge Rey’s Python Spatial Analysis Library (PySAL) (Rey
et al., 2015; Rey, 2019). Paul Waddell’s UrbanSim platform, a Python-based software library
that supports urban land use and transportation modeling and simulation, is similarly free
and open-source (Waddell, 2002). Beyond this sort of hero model of tool-building, many other
community-driven open-source spatial software projects exist in the Python ecosystem and
other programming languages (e.g., GDAL, geopandas, leaflet, PostGIS, QGIS, r-spatial, and
even OpenStreetMap itself). Such projects often utilize a decentralized many hands model
to grow a community around building better tools and collaborating synergistically around
shared goals. Regardless of the exact model, academic-led projects can provide a theory-rich
scaffolding on which research can be conducted and other tools (like OSMnx) can be built.
These tools and their ongoing development are also important for empowering students and
scholars who cannot afford expensive proprietary software licenses. We must reduce such
barriers to entry.
Academics too often pull up the gangplanks by placing scientific findings behind paywalls,
hoarding useful datasets, or concealing their research software. The intertwined open-science,
open-data, and open-source movements address these three respectively by publicly sharing
scientific findings, data, and software for the good of society. This in turn disseminates knowl-
edge and empowers the wider scientific community. If our goal as academics is to produce
empirical research and advance theory, our time should focus on research and writing—but
we should also set aside time to build better theory-rich tools to answer difficult questions.
5.3. Broader Community Impacts
Third, open-source tool-building impacts a broad community with bidirectional effects as
the tools we build eventually underpin others’ work downstream, while other researchers
contribute back to our projects upstream. As two examples of downstream effects, the
mobile crowd-sensing platform CrowdSenSim now uses OSMnx to simulate urban environ-
ments (Tomasoni et al., 2018; Montori et al., 2019) and the transportation planning company
Remix developed its platform—now deployed in hundreds of cities worldwide—initially
using OSMnx to model street networks. Reciprocally, regarding the upstream direction, as an
open-source project OSMnx has received hundreds of code contributions from other scholars
and members of the public. These contributors helped develop its points-of-interest module,
its nearest-node and nearest-edge search algorithms, its building footprint functionality, and
much more. I have reaped the benefits of dozens of others’ contributions that enhanced the
tool in ways I subsequently used to answer my own research questions.
12
Better tools and data models, spearheaded by academics, can help infuse theory into our
field’s quantitative work. But if we want better tools, we have to build them. It is not ESRI’s
job to satisfy all the theoretical needs of the spatial sciences. Recent important scholarly
critiques have highlighted how today’s GIScience tool landscape is inaccessible, atheoretical,
and ad hoc (e.g., Gahegan, 2018; Poorthuis & Zook, 2019). One clear path to better link this
academic critique with tangible real-world action is to build and incentivize better tools for
praxis. Collectively, we need to spend more time fixing the lack of high-quality accessible
tools rather than just writing yet another thinkpiece lamenting their dearth.
6. Toward Better Tools
Software often feels inevitable because its backstory is often invisible. We click a download
link, run an installer, and suddenly have a new tool to use. Yet this conceals years of human
decisions, experiences, and constraints shaping software outcomes that are in no way pre-
destined. In many ways, open-source software parallels other public infrastructure such as
highways and bridges. Humans plan and engineer infrastructure in specific social, economic,
and technological contexts. Like an individual highway’s or bridge’s broader network of
connected roads, any single piece of software represents a complex network of entangle-
ments with many other software projects on which it depends. And like a highway or bridge,
software requires years of maintenance, updates, and retrofits after its initial development:
although splashy new capital projects often receive funding and adulation, critical routine
maintenance work usually receives far less of either.
Tool-building, in all its facets, remains an essential but poorly incentivized pillar of
academia today. Computational geographic tools too often impoverish quantitative analytics
through poor representations of scientific theory and squander precious time as everyone
develops their own ad hoc scripts to solve similar problems. The case of urban street network
science illustrated this. Building an open-source, reusable, accessible, theoretically-sound
tool as public infrastructure has generated various downstream and upstream benefits. So
what can we do to foster a robust tool-building community and collectively reap more of
these benefits?
Incentives are key. First, academic tenure, promotion, and annual review guidelines
should explicitly reward the scholarly value of open-source research software and open data
contributions to better acknowledge their significant value. They should also account for
contributions to pre-existing and decentralized projects to encourage collaborative progress
and maintenance of the open-source commons. An ideal system would better balance the
respective value of research, publication, and tool-building to advancing science and theory.
Second, we should train the next generation of practitioners and scholars to be better tool cre-
ators and consumers. Curricula should include coding and informatics courses and pedagogy
should emphasize hands-on learning, such as using computational notebooks. Third, aca-
demic publication must continue its nascent steps toward open science: journal editors should
require the submission and publication of datasets and computational workflows alongside
13
quantitative manuscripts. Finally, we should increase funding opportunities for building and
maintaining research software to foster the positive externalities and downstream benefits
they generate throughout the research community.
The goal of academic tool-building—and the hope of every such tool-builder—is to
construct some kind of useful infrastructure for your field. With better scaffolding in place to
connect theory, science, and analytics, we can conduct better research to explore important
geospatial questions and foreground objects, relations, and processes. OSMnx has not changed
the world or reinvented urban science, but it has made empirical street network analysis a
little easier and more reproducible for some urban scholars and practitioners. Hopefully it
even unlocked a study or two along the way that otherwise could not have been conducted.
Finally, there is one key lesson-learned that I would like to share with other budding academic
tool-builders: give your tool an easier name to remember and pronounce than OSMnx.
Acknowledgments
This work was funded in part by a grant from The Public Good Projects. I wish to thank
the editors of Transactions in GIS for the opportunity to present this plenary paper and two
anonymous reviewers for their helpful suggestions in improving it. I also wish to thank
the innumerable other developers of open-source software on which my research and tool-
building rely. Figure 1 is reprinted with permission from Boeing (2019a) and Figure 2 is
reprinted from Boeing (2019b) under the terms of the Creative Commons Attribution 4.0
International License.
Notes
1
A simple example of these objects and relations would be: Los Angeles
→
is a
→
city; USC
→
is a
→
university; USC →is in →Los Angeles
2
The “rule of three” in computer programming states that if similar code will be used in three or more places,
it should be extracted for general reuse to avoid duplication. We might consider similar guidelines in making our
research tools publicly reusable.
3
OSMnx installation instructions and user documentation are available online at https://osmnx.readthedocs.org/
and the open-source project itself is hosted at https://github.com/gboeing/osmnx
4
OSMnx tutorials and usage examples in Jupyter notebook format are available online at https://github.com/gboeing/osmnx-
examples
References
Abdelkader, A., Boeing, G., Fasy, B. T., & Millman, D. L. (2018). Topological Distance
Between Nonplanar Transportation Networks. In 28th Fall Workshop on Computational
Geometry (pp. 1–6). Queens, NY.
Acuto, M., Parnell, S., & Seto, K. C. (2018). Building a Global Urban Science. Nature
Sustainability,1(1), 2–4. https://doi.org/10.1038/s41893-017-0013-9
14
Adler, S. O., Bodeit, O., Bonn, L., Goldenbogen, B., Haffner, J. E. L., Karnetzki, M., .. . Klipp,
E. (2020). Geospatially Referenced Demographic Agent-Based Modeling of SARS-CoV-
2-Infection (COVID-19) Dynamics and Mitigation Effects in a Real-world Community.
(medRxiv preprint). https://doi.org/10.1101/2020.05.03.20089235
Alberti, M. (2017). Grand Challenges in Urban Science. Frontiers in Built Environment,3(6),
1–5. https://doi.org/10.3389/fbuil.2017.00006
Anselin, L., Syabri, I., & Kho, Y. (2006). GeoDa: An Introduction to Spatial Data Analysis.
Geographical Analysis,38(1), 5–22. https://doi.org/10.1111/j.0016-7363.2005.00671.x
Arribas-Bel, D., & Reades, J. (2018). Geography and Computers: Past, Present, and Future.
Geography Compass,12(10), e12403. https://doi.org/10.1111/gec3.12403
Barron, C., Neis, P., & Zipf, A. (2014). A Comprehensive Framework for In-
trinsic OpenStreetMap Quality Analysis. Transactions in GIS,18(6), 877–895.
https://doi.org/10.1111/tgis.12073
Barthelemy, M. (2011). Spatial Networks. Physics Reports,499(1-3), 1–101.
https://doi.org/10.1016/j.physrep.2010.11.002
Barthelemy, M. (2019). The Statistical Physics of Cities. Nature Reviews Physics,1(6),
406–415. https://doi.org/10.1038/s42254-019-0054-2
Basiri, A., Jackson, M., Amirian, P., Pourabdollah, A., Sester, M., Winstanley, A., ... Zhang, L.
(2016). Quality assessment of OpenStreetMap data using trajectory mining. Geospatial
Information Science,19(1), 56–68. https://doi.org/10.1080/10095020.2016.1151213
Batty, M. (1971). Modelling Cities as Dynamic Systems. Nature,231(5303), 425–428.
https://doi.org/10.1038/231425a0
Batty, M. (1980). Limits to Prediction in Science and Design Science. Design Studies,1(3),
153–159. https://doi.org/10.1016/0142-694X(80)90022-8
Batty, M. (2013). The New Science of Cities. Cambridge, MA: MIT Press.
Batty, M. (2019). Urban Analytics Defined. Environment and Planning B: Urban Analytics and
City Science,46(3), 403–405. https://doi.org/10.1177/2399808319839494
Baumann, P., & Keupp, M. M. (2020). Assessing the Reliability of Street Networks: A Case
Study Based on the Swiss Street Network. In M. M. Keupp (Ed.), The Security of Critical
Infrastructures: Risk, Resilience and Defense (pp. 111–129). Cham, Switzerland: Springer.
https://doi.org/10.1007/978-3-030-41826-7_8
Behrend, L., & Levin-Keitel, M. (2019). Planning as Scientific Discipline? Digging
Deep toward the Bottom Line of the Debate. Planning Theory, 147309521989728.
https://doi.org/10.1177/1473095219897283
Beirigo, B., Schulte, F., & Negenborn, R. (2018). Dual-Mode Vehicle Routing in Mixed
Autonomous and Non-Autonomous Zone Networks. In 2018 21st International Con-
ference on Intelligent Transportation Systems (ITSC) (pp. 1325–1330). Maui, HI: IEEE.
https://doi.org/10.1109/ITSC.2018.8569344
Bertuglia, C. S., Bianchi, G., & Mela, A. (Eds.). (1998). The City and Its Sciences. Heidelberg,
Germany: Physica-Verlag. https://doi.org/10.1007/978-3-642-95929-5
15
Boeing, G. (2017). OSMnx: New Methods for Acquiring, Constructing, Analyzing, and
Visualizing Complex Street Networks. Computers, Environment and Urban Systems,65,
126–139. https://doi.org/10.1016/j.compenvurbsys.2017.05.004
Boeing, G. (2018). Planarity and Street Network Representation in Urban Form
Analysis. Environment and Planning B: Urban Analytics and City Science(online first).
https://doi.org/10.1177/2399808318802941
Boeing, G. (2019a). The Morphology and Circuity of Walkable and Drivable Street Net-
works. In L. D’Acci (Ed.), The Mathematics of Urban Morphology (pp. 271–287). Basel,
Switzerland: Birkhäuser. https://doi.org/10.1007/978-3-030-12381-9_12
Boeing, G. (2019b). Urban Spatial Order: Street Network Orientation, Configuration, and
Entropy. Applied Network Science,4(1), 67. https://doi.org/10.1007/s41109-019-0189-1
Boeing, G. (2020a). A Multi-Scale Analysis of 27,000 Urban Street Networks: Every US City,
Town, Urbanized Area, and Zillow Neighborhood. Environment and Planning B: Urban
Analytics and City Science,47(4), 590–608. https://doi.org/10.1177/2399808318784595
Boeing, G. (2020b). Spatial Information and the Legibility of Urban Form: Big Data
in Urban Morphology. International Journal of Information Management(online first).
https://doi.org/10.1016/j.ijinfomgt.2019.09.009
Branch, M. C. (1966). Simulation, Mathematical Models, and Comprehensive City Planning.
Urban Affairs Quarterly,1(3), 15–38. https://doi.org/10.1177/107808746600100302
Brandily, P., & Rauch, F. (2018). Roads and Urban Growth (Department of Economics Paper
No. 859). Oxford, England: University of Oxford.
Bristow, N. J. O. (2019). Measuring residential building density at the scale of the urban block
(Thesis). University of Westminster, London, England.
Brunsdon, C. (2016). Quantitative Methods I: Reproducible Research and
Quantitative Geography. Progress in Human Geography,40(5), 687–696.
https://doi.org/10.1177/0309132515599625
Burgess, E. W. (1925). Can Neighborhood Work Have a Scientific Basis? In R. E. Park &
E. W. Burgess (Eds.), The City (pp. 142–155). Chicago, IL: University of Chicago Press.
Corcoran, P., Mooney, P., & Bertolotto, M. (2013). Analysing the growth of OpenStreetMap
networks. Spatial Statistics,3, 21–32. https://doi.org/10.1016/j.spasta.2013.01.002
Coutrot, A., Manley, E., Yesiltepe, D., Dalton, R., Wiener, J. M., Hölscher, C., .. . Spiers, H. J.
(2020). Cities Have a Negative Impact on Navigation Ability: Evidence from 38 Countries.
(bioRxiv preprint). https://doi.org/10.1101/2020.01.23.917211
D’Angelo, G., Ferretti, S., & Ghini, V. (2017). Modeling the Internet of Things: a simulation
perspective. In 2017 International Conference on High Performance Computing & Simulation
(HPCS) (pp. 18–27). Genoa, Italy: IEEE. https://doi.org/10.1109/HPCS.2017.13
Derudder, B., & van Meeteren, M. (2019). Engaging with "Urban Science". Urban Geogra-
phy(online first), 1–10. https://doi.org/10.1080/02723638.2019.1585138
Dingil, A. E., Rupi, F., & Stasiskiene, Z. (2019). A Macroscopic analysis of trans-
port networks: The influence of network design on urban transportation perfor-
16
mance. International Journal of Transport Development and Integration,3(4), 331–343.
https://doi.org/10.2495/TDI-V3-N4-331-343
Donoho, D. (2017). 50 Years of Data Science. Journal of Computational and Graphical Statistics,
26(4), 745–766. https://doi.org/10.1080/10618600.2017.1384734
Dumedah, G., & Eshun, G. (2020). The Case of Paratransit: Trotro Service Data as a Credible
Location Addressing of Road Networks in Ghana. Journal of Transport Geography,84,
102688. https://doi.org/10.1016/j.jtrangeo.2020.102688
Feng, M., & Porter, M. A. (2020). Spatial Applications of Topological Data Analysis:
Cities, Snowflakes, Random Structures, and Spiders Spinning under the Influence.
(arXiv:2001.01872). Retrieved from http://arxiv.org/abs/2001.01872
Gahegan, M. (1999). What Is Geocomputation? Transactions in GIS,3(3), 203–206.
https://doi.org/10.1111/1467-9671.00017
Gahegan, M. (2018). Our GIS Is Too Small. The Canadian Geographer,62(1), 15–26.
https://doi.org/10.1111/cag.12434
Gahegan, M. (2020). Fourth Paradigm GIScience? Prospects for Automated Discovery and
Explanation from Data. International Journal of Geographical Information Science,34(1),
1–21. https://doi.org/10.1080/13658816.2019.1652304
Geospatial Software Institute. (2020). GSI: Geospatial Software Institute. Retrieved 2020-05-
20, from https://gsi.cigi.illinois.edu/
Gervasoni, L., Bosch, M., Fenet, S., & Sturm, P. (2017). Calculating spatial urban sprawl
indices using open data. In 15th International Conference on Computers in Urban Planning
and Urban Management. Adelaide, Australia.
Girres, J.-F., & Touya, G. (2010). Quality Assessment of the French OpenStreetMap Dataset.
Transactions in GIS,14(4), 435–459. https://doi.org/10.1111/j.1467-9671.2010.01203.x
Graham, M., De Sabbata, S., & Zook, M. A. (2015). Towards a study of information ge-
ographies: (im)mutable augmentations and a mapping of the geographies of information:
Towards a study of information geographies. Geo: Geography and Environment,2(1),
88–105. https://doi.org/10.1002/geo2.8
Haklay, M. (2010). How Good is Volunteered Geographical Information? A Comparative
Study of OpenStreetMap and Ordnance Survey Datasets. Environment and Planning B:
Planning and Design,37(4), 682–703. https://doi.org/10.1068/b35097
Harris, R., O’Sullivan, D., Gahegan, M., Charlton, M., Comber, L., Longley, P., .. . Evans,
A. (2017). More Bark than Bytes? Reflections on 21+ Years of Geocomputa-
tion. Environment and Planning B: Urban Analytics and City Science,44(4), 598–617.
https://doi.org/10.1177/2399808317710132
Hellervik, A., Nilsson, L., & Andersson, C. (2019). Preferential centrality: A new measure uni-
fying urban activity, attraction and accessibility. Environment and Planning B: Urban Ana-
lytics and City Science,46(7), 1331–1346. https://doi.org/10.1177/2399808318812888
Hernández-Hernández, A. M., Siqueiros-García, J. M., Robles-Belmont, E., & Gershen-
son, C. (2019). Anger while driving in Mexico City. PLOS One,14(9), e0223048.
https://doi.org/10.1371/journal.pone.0223048
17
Hofer, C., Jäger, G., & Füllsack, M. (2018a). Including traffic jam avoidance in an agent-based
network model. Computational Social Networks,5(1), 5. https://doi.org/10.1186/s40649-
018-0053-y
Hofer, C., Jäger, G., & Füllsack, M. (2018b). Large scale simulation of CO2 emissions caused
by urban car traffic: An agent-based network approach. Journal of Cleaner Production,183,
1–10. https://doi.org/10.1016/j.jclepro.2018.02.113
Holub, T. (2017). The Invisible Messenger Bag: Cycling Advocacy and Identity Politics in Selected
U.S. Cities (Thesis). University of California, Berkeley, Berkeley, CA.
Hoyt, H. (1951). Is City Growth Controlled by Mathematics or Physical Laws? Land
Economics,27(3), 259. https://doi.org/10.2307/3159384
Johannesen, J.-A., Olaisen, J., & Olsen, R. (1998). The Philosophy of Science, Planning and
Decision Theories. Built Environment,24(2/3), 155–168.
Jones, N. (2019). Identifying gaps in OpenStreetMap coverage through machine learning.
Retrieved from
https://towardsdatascience.com/identifying-gaps-in
-openstreetmap-coverage-through-machine-learning-257545c04330
Kang, J.-Y., Michels, A. C., Lyu, F., Wang, S., Agbodo, N., Freeman, V. L., &
Wang, S. (2020). Rapidly Measuring Spatial Accessibility of COVID-
19 Healthcare Resources: A Case Study of Illinois, USA. (medRxiv preprint).
https://doi.org/10.1101/2020.05.06.20093534
Kang, W., Oshan, T., Wolf, L. J., Boeing, G., Frias-Martinez, V., Gao, S., . .. Xu,
W. (2019). A Roundtable Discussion: Defining Urban Data Science. En-
vironment and Planning B: Urban Analytics and City Science,46(9), 1756–1768.
https://doi.org/10.1177/2399808319882826
Kedron, P., Frazier, A. E., Trgovac, A. B., Nelson, T., & Fotheringham, A. S. (2019). Repro-
ducibility and Replicability in Geographical Analysis. Geographical Analysis(online first).
https://doi.org/10.1111/gean.12221
Kitchin, R. (2016). The Ethics of Smart Cities and Urban Science. Philosophical Transactions of
the Royal Society A: Mathematical, Physical and Engineering Sciences,374(2083), 20160115.
https://doi.org/10.1098/rsta.2016.0115
Kontokosta, C. E. (2018). Urban Informatics in the Science and Practice of Plan-
ning. Journal of Planning Education and Research(online first), 0739456X1879371.
https://doi.org/10.1177/0739456X18793716
Koster, S., & Rowe, F. (2020). Fueling Research Transparency: Com-
putational Notebooks and the Discussion Section. Region,6(3), E1–E2.
https://doi.org/10.18335/region.v6i3.309
Law, S., & Neira, M. (2019). An unsupervised approach to Geographical Knowledge
Discovery using street level and street network images. (arXiv:1906.11907). Retrieved
from http://arxiv.org/abs/1906.11907
Lee, D. B. (1973). Requiem for Large-Scale Models. Journal of the American Institute of
Planners,39(3), 163–178. https://doi.org/10.1080/01944367308977851
18
Lee, D. B. (1994). Retrospective on Large-Scale Urban Models. Journal of the American
Planning Association,60(1), 35–40. https://doi.org/10.1080/01944369408975549
Liao, Y., Gil, J., Pereira, R. H. M., Yeh, S., & Verendel, V. (2020). Disparities in travel times
between car and transit: Spatiotemporal patterns in cities. Scientific Reports,10(1), 4056.
https://doi.org/10.1038/s41598-020-61077-0
Lin, D., Zhang, Y., Zhu, R., & Meng, L. (2019). The analysis of catchment areas of metro
stations using trajectory data generated by dockless shared bikes. Sustainable Cities and
Society,49, 101598. https://doi.org/10.1016/j.scs.2019.101598
Lin, Q., Deng, L., Sun, J., & Chen, M. (2018). Optimal Demand-Aware Ride-Sharing
Routing. In IEEE INFOCOM 2018 (pp. 2699–2707). Honolulu, Hawaii: IEEE.
https://doi.org/10.1109/INFOCOM.2018.8486278
Liu, Y., Zhang, Y., Jin, S. T., & Liu, Y. (2020). Spatial pattern of leisure activities among
residents in Beijing, China: Exploring the impacts of urban environment. Sustainable
Cities and Society,52, 101806. https://doi.org/10.1016/j.scs.2019.101806
Lobo, J., Alberti, M., Allen-Dumas, M., Arcaute, E., Barthelemy, M., Tapia, B., .. . Youn, H.
(2020). Urban Science: Integrated Theory from the First Cities to Sustainable Metropolises
(Mansueto Institute for Urban Innovation Research Paper). Chicago, IL: University of
Chicago. Retrieved from https://doi.org/10.2139/ssrn.3526940
Luo, M., Du, B., Klemmer, K., Zhu, H., Ferhatosmanoglu, H., & Wen, H. (2020). D3P:
Data-driven Demand Prediction for Fast Expanding Electric Vehicle Sharing Systems.
Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies,4(1),
1–21. https://doi.org/10.1145/3381005
Maier, G. (2014). OpenStreetMap, the Wikipedia Map. Region,1(1), R3–R10.
https://doi.org/10.18335/region.v1i1.70
Marshall, S., Gil, J., Kropf, K., Tomko, M., & Figueiredo, L. (2018). Street Network Studies:
from Networks to Models and their Representations. Networks and Spatial Economics.
https://doi.org/10.1007/s11067-018-9427-9
Martínez Mori, J. C., & Samaranayake, S. (2019). Bounded Asymmetry in Road Networks.
Scientific Reports,9(1), 11951. https://doi.org/10.1038/s41598-019-48463-z
Masias, V. H., Hecking, T., Crespo, F., & Hoppe, H. U. (2019). Detecting social media users
based on pedestrian networks and neighborhood attributes: an observational study.
Applied Network Science,4(1), 96. https://doi.org/10.1007/s41109-019-0222-4
Mattern, S. (2013). Methodolatry and the Art of Measure. Places(November).
https://doi.org/10.22269/131105
Merchán, D., & Winkenbach, M. (2019). An empirical validation and data-driven extension
of continuum approximation approaches for urban route distances. Networks,73(4),
418–433. https://doi.org/10.1002/net.21874
Merchán, D., Winkenbach, M., & Snoeck, A. (2020). Quantifying the impact of urban road
networks on the efficiency of local trips. Transportation Research Part A: Policy and Practice,
135, 38–62. https://doi.org/10.1016/j.tra.2020.02.015
19
Montori, F., Cortesi, E., Bedogni, L., Capponi, A., Fiandrino, C., & Bononi, L. (2019). Crowd-
SenSim 2.0: A Stateful Simulation Platform for Mobile Crowdsensing in Smart Cities. In
Proceedings of the 22nd International ACM Conference on Modeling, Analysis and Simulation
of Wireless and Mobile Systems - MSWIM ’19 (pp. 289–296). Miami Beach, FL: ACM Press.
https://doi.org/10.1145/3345768.3355929
Morelli, A. B., & Cunha, A. L. (2019). Verificação de Vulnerabilidades em Redes de Trans-
porte. In Congresso de Pesquisa e Ensino em Transportes da ANPET. Balneário Camboriú,
Brazil.
Natera Orozco, L. G., Battiston, F., Iñiguez, G., & Szell, M. (2019). Data-Driven Strategies
for Optimal Bicycle Network Growth. (arXiv:1907.07080). Retrieved from
http://
arxiv.org/abs/1907.07080
Natera Orozco, L. G., Deritei, D., Vancso, A., & Vasarhelyi, O. (2020). Quantifying Life
Quality as Walkability on Urban Networks: The Case of Budapest. In H. Cherifi, S. Gaito,
J. F. Mendes, E. Moro, & L. M. Rocha (Eds.), Complex Networks and Their Applications VIII
(pp. 905–918). Cham, Switzerland: Springer.
Neukart, F., Compostella, G., Seidel, C., von Dollen, D., Yarkoni, S., & Parney, B. (2017).
Traffic Flow Optimization Using a Quantum Annealer. Frontiers in ICT,4, 29.
https://doi.org/10.3389/fict.2017.00029
O’Keeffe, K., Santi, P., & Ratti, C. (2019). Modeling vehicular mobility patterns using
recurrent neural networks. (arXiv:1910.11851). Retrieved from
http://arxiv.org/
abs/1910.11851
O’Sullivan, D., & Manson, S. M. (2015). Do Physicists Have Geography Envy? And What
Can Geographers Learn from It? Annals of the Association of American Geographers,105(4),
704–722. https://doi.org/10.1080/00045608.2015.1039105
Padgham, M., Boeing, G., Cooley, D., Tierney, N., Sumner, M., Phan, T. G., & Beare, R. (2019).
An Introduction to Software Tools, Data, and Services for Geospatial Analysis of Stroke
Services. Frontiers in Neurology,10, 743. https://doi.org/10.3389/fneur.2019.00743
Poorthuis, A., & Zook, M. (2019). Being Smarter about Space: Drawing Lessons from
Spatial Science. Annals of the American Association of Geographers(online first), 1–11.
https://doi.org/10.1080/24694452.2019.1674630
Quistberg, D. A., Diez Roux, A. V., Bilal, U., Moore, K., Ortigoza, A., Rodriguez, D. A.,
.. . Miranda, J. J. (2019). Building a Data Platform for Cross-Country Urban
Health Studies: the SALURBAL Study. Journal of Urban Health,96(2), 311–337.
https://doi.org/10.1007/s11524-018-00326-0
Ren, Y., Cheng, T., & Zhang, Y. (2019). Deep spatio-temporal residual neural networks
for road-network-based data modeling. International Journal of Geographical Information
Science,33(9), 1894–1912. https://doi.org/10.1080/13658816.2019.1599895
Rey, S. J. (2009). Show Me the Code: Spatial Analysis and Open Source. Journal of Geographi-
cal Systems,11(2), 191–207. https://doi.org/10.1007/s10109-009-0086-8
Rey, S. J. (2019). PySAL: The First 10 Years. Spatial Economic Analysis,14(3), 273–282.
https://doi.org/10.1080/17421772.2019.1593495
20
Rey, S. J., Anselin, L., Li, X., Pahle, R., Laura, J., Li, W., & Koschinsky, J. (2015). Open
Geospatial Analytics with PySAL. ISPRS International Journal of Geo-Information,4(2),
815–836. https://doi.org/10.3390/ijgi4020815
Riascos, A. P., & Mateos, J. L. (2020). Networks and long-range mobility in cities: A study
of more than one billion taxi trips in New York City. Scientific Reports,10(1), 4022.
https://doi.org/10.1038/s41598-020-60875-w
Saha, S. S., Schweitzer, E., Scaglione, A., & Johnson, N. G. (2019). A Framework for Generat-
ing Synthetic Distribution Feeders using OpenStreetMap. (arXiv:1910.07673). Retrieved
from http://arxiv.org/abs/1910.07673
Sallis, J. F., Bull, F., Burdett, R., Frank, L. D., Griffiths, P., Giles-Corti, B., & Stevenson,
M. (2016). Use of Science to Guide City Planning, Policy, and Practice: How to
Achieve Healthy and Sustainable Future Cities. The Lancet,388(10062), 2936–2947.
https://doi.org/10.1016/S0140-6736(16)30068-X
Samson, B. P. V., Velez, G. A. T., Nobleza, J. R., Sanchez, D., & Milan, J. T. (2018). Optimizing
the Efficiency, Vulnerability and Robustness of Road-Based Para-Transit Networks
Using Genetic Algorithm. In Y. Shi et al. (Eds.), Computational Science – ICCS 2018 (Vol.
10860, pp. 3–14). Cham, Switzerland: Springer.
Schmidt, M., & Klettner, S. (2013). Gender and Experience-Related Motivators for Con-
tributing to OpenStreetMap. In AGILE 2013. Leuven, Belgium: AGILE.
Sehra, S. S., Singh, J., Rai, H. S., & Anand, S. S. (2019). Extending Processing Toolbox for
Assessing the Logical Consistency of OpenStreetMap Data. Transactions in GIS(online
first), 1–28. https://doi.org/10.1111/tgis.12587
Senturk, I. F., & Kebe, G. Y. (2019). A Novel Shortest Path Routing Algorithm for Wire-
less Data Collection in Transportation Networks. In 2019 4th International Confer-
ence on Computer Science and Engineering (UBMK) (pp. 1–5). Samsun, Turkey: IEEE.
https://doi.org/10.1109/UBMK.2019.8907167
Singleton, A., & Arribas-Bel, D. (2019). Geographic Data Science. Geographical Analy-
sis(online first). https://doi.org/10.1111/gean.12194
Sohouenou, P. Y. R., Christidis, P., Christodoulou, A., Neves, L. A. C., & Presti, D. L.
(2020). Using a random road graph model to understand road networks robustness
to link failures. International Journal of Critical Infrastructure Protection,29, 100353.
https://doi.org/10.1016/j.ijcip.2020.100353
Solecki, W., Seto, K. C., & Marcotullio, P. J. (2013). It’s Time for an Urbanization
Science. Environment: Science and Policy for Sustainable Development,55(1), 12–17.
https://doi.org/10.1080/00139157.2013.748387
Tomasoni, M., Capponi, A., Fiandrino, C., Kliazovich, D., Granelli, F., & Bouvry, P. (2018).
Profiling Energy Efficiency of Mobile Crowdsensing Data Collection Frameworks
for Smart City Applications. In 2018 6th IEEE International Conference on Mobile
Cloud Computing, Services, and Engineering (MobileCloud) (pp. 1–8). Bamberg: IEEE.
https://doi.org/10.1109/MobileCloud.2018.00009
21
Torres, J. E. H., González, S. H., García, J. A. J., & Fernández, V. F. (2019). Análisis de
Vulnerabilidad de la Infraestructura de Transporte Aplicando Redes Complejas: Red de
Avenidas de la Ciudad de Celaya, Guanajuato. Pistas Educativas,41(133).
Van Etten, A., Lindenbaum, D., & Bacastow, T. M. (2019). SpaceNet: A Remote Sensing
Dataset and Challenge Series. (arXiv:1807.01232). Retrieved from
http://arxiv.org/
abs/1807.01232
Verendel, V., & Yeh, S. (2019). Measuring Traffic in Cities Through a Large-Scale
Online Platform. Journal of Big Data Analytics in Transportation,1(2), 161–173.
https://doi.org/10.1007/s42421-019-00007-7
Waddell, P. (2002). UrbanSim: Modeling Urban Development for Land Use, Transportation,
and Environmental Planning. Journal of the American Planning Association,68(3), 297–314.
https://doi.org/10.1080/01944360208976274
Waddell, P., Garcia-Dorado, I., Maurer, S. M., Boeing, G., Gardner, M., Porter, E., & Aliaga,
D. (2018). Architecture for Modular Microsimulation of Real Estate Markets and
Transportation. In Applied Urban Modelling Symposium. Cambridge, England. Retrieved
from https://arxiv.org/pdf/1807.01148
Wang, M., Chen, Z., Mu, L., & Zhang, X. (2020). Road network structure and ride-sharing
accessibility: A network science perspective. Computers, Environment and Urban Systems,
80, 101430. https://doi.org/10.1016/j.compenvurbsys.2019.101430
Wang, S., Gao, S., Feng, X., Murray, A. T., & Zeng, Y. (2018). A context-based geoprocess-
ing framework for optimizing meetup location of multiple moving objects along road
networks. International Journal of Geographical Information Science,32(7), 1368–1390.
https://doi.org/10.1080/13658816.2018.1431838
Wang, S., Xu, G., & Guo, Q. (2018). Street Centralities and Land Use Intensities Based on
Points of Interest (POI) in Shenzhen, China. ISPRS International Journal of Geo-Information,
7(11), 425. https://doi.org/10.3390/ijgi7110425
Wang, S., Yu, D., Kwan, M.-P., Zheng, L., Miao, H., & Li, Y. (2020). The Impacts of Road
Network Density on Motor Vehicle Travel: An Empirical Study of Chinese Cities Based
on Network Theory. Transportation Research Part A: Policy and Practice,132, 144–156.
https://doi.org/10.1016/j.tra.2019.11.012
Yan, B., Janowicz, K., Mai, G., & Zhu, R. (2019). A Spatially Explicit Reinforcement Learning
Model for Geographic Knowledge Graph Summarization. Transactions in GIS,23(3),
620–640. https://doi.org/10.1111/tgis.12547
Yin, Y., Varadarajan, J., Wang, G., Wang, X., Sahrawat, D., Zimmermann, R., & Ng, S.-
K. (2020). A Multi-task Learning Framework for Road Attribute Updating via
Joint Analysis of Map Data and GPS Traces. In Proceedings of The Web Conference
2020 (pp. 2662–2668). Taipei, Taiwan: Association for Computing Machinery.
https://doi.org/10.1145/3366423.3380021
Young, D. L., & Eccles, C. (2020). Automatic construction of Markov decision process mod-
els for multi-agent reinforcement learning. In Artificial Intelligence and Machine Learning
22
for Multi-Domain Operations Applications II (Vol. 11413, p. 114130Y). International Society
for Optics and Photonics. https://doi.org/10.1117/12.2557823
Zhang, Y., Lin, D., & Mi, Z. (2019). Electric fence planning for dock-
less bike-sharing services. Journal of Cleaner Production,206, 383–393.
https://doi.org/10.1016/j.jclepro.2018.09.215
Zielstra, D., Hochmair, H. H., & Neis, P. (2013). Assessing the Effect of Data Imports on the
Completeness of OpenStreetMap – A United States Case Study. Transactions in GIS,17(3),
315–334. https://doi.org/10.1111/tgis.12037
23
