Friday, January 20, 2017

Authoritative and VGI in a Developing Country: A Comparative Case Study of Road Datasets in Nairobi

The motivation behind the paper was that while there are numerous studies comparing VGI to authoritative data in the developed world, there are very few that do so in developing world. In order to address this issue in the paper we compare the quality of authoritative road data (i.e. from the Regional Center for Mapping of Resources for Development - RCMRD) and non-authoritative crowdsourced road data (i.e. from OpenStreetMap (OSM) and Google’s Map Maker) in conjunction with population data in and around Nairobi, Kenya.

Results from our analysis show variability in coverage between all these datasets. RCMRD provided the most complete, albeit less current, coverage when taking into account the entire study area, while OSM and Map Maker showed a degradation of coverage as one moves from central Nairobi towards more rural areas. Further information including the abstract to our paper, some figures and full reference is given below.

With volunteered geographic information (VGI) platforms such as OpenStreetMap (OSM) becoming increasingly popular, we are faced with the challenge of assessing the quality of their content, in order to better understand its place relative to the authoritative content of more traditional sources. Until now, studies have focused primarily on developed countries, showing that VGI content can match or even surpass the quality of authoritative sources, with very few studies in developing countries. In this paper we compare the quality of authoritative (data from the Regional Center for Mapping of Resources for Development - RCMRD) and non-authoritative (data from OSM and Google’s Map Maker) road data in conjunction with population data in and around Nairobi, Kenya. Results show variability in coverage between all these datasets. RCMRD provided the most complete, albeit less current, coverage when taking into account the entire study area, while OSM and Map Maker showed a degradation of coverage as one moves from central Nairobi towards rural areas. Furthermore, OSM had higher content density in large slums, surpassing the authoritative datasets at these locations, while Map Maker showed better coverage in rural housing areas. These results suggest a greater need for a more inclusive approach using VGI to supplement gaps in authoritative data in developing nations.

Keywords: Volunteered Geographic Information; Crowdsourcing; Road Networks; Population Data; Kenya  
Road Coverage per km2
Pairwise difference in road coverage. Clockwise from top left: i) RCMRD 2011 versus Map Maker 2014; ii) RCMRD 2011 versus OSM 2011; iii) RCMRD 2011 versus OSM 2014; iv) OSM 2014 versus Map Maker 2014 (Red cells: first layer has higher coverage; Green cells: second layer has higher coverage).

Full Reference:
Mahabir, R., Stefanidis, A., Croitoru, A., Crooks, A.T. and Agouris, P. (2017), “Authoritative and Volunteered Geographical Information in a Developing Country: A Comparative Case Study of Road Datasets in Nairobi, Kenya”, ISPRS International Journal of Geo-Information, 6(1): 24, doi:10.3390/ijgi6010024.
As always any thoughts or comments about this work are welcome.

Thursday, January 12, 2017

Transportation in Agent-Based Urban Modelling

Sarah Wise, Mike Batty and myself have recently had a chapter published in Agent Based Modelling of Urban Systems entitled "Transportation in Agent-Based Urban Modelling". In the chapter we provide a critique in how transportation has been included or omitted from agent-based models and suggest how it might be handled in future applications.

Our argument is that transportation plays an important role in nearly every aspect of our daily lives. However, within agent-based models that explore urban problems, transportation is often omitted. Using representative case studies (e.g. from crime, disease spread, and land use) we present different levels/tiers of complexity at which transportation systems are captured with agent-based models (as shown in Table 1). Table 2 shows how these tiers of complexity are captured within crime models.  For interested readers, below you can see the abstract to our chapter.  

As the urban population rapidly increases to the point where most of us will be living in cities by the end of this century, the need to better understand urban areas grows ever more urgent. Urban simulation modelling as a field has developed in response to this need, utilizing developing technologies to explore the complex inter-dependencies, feedback's, and heterogeneities which characterize and drive processes that link the functions of urban areas to their form. As these models grow more nuanced and powerful, it is important to consider the role of transportation within them. Transportation joins, divides, and structures urban areas, providing a functional definition of the geometry and the economic costs that determine urban processes accordingly. However, it has proved challenging to factor transportation into agent-based models (ABM); past approaches to such modelling have struggled to incorporate information about accessibility, demographics, or time costs in a significant way. ABM have not yet embraced alternative traditions such as that in land use transportation modelling that build on spatial interaction in terms of transport directly, nor have these alternate approaches been disaggregated to the level at which populations are represented as relatively autonomous agents. Where disaggregation of aggregate transport has taken place, it has led to econometric models of individual choice or microsimulaton models of household activity patterns which only superficially embody the key principles of ABM. But the explosion in the availability of movement data in recent years, combined with improvements in modelling technology, is easing this process dramatically. In particular, agent-based modelling as a methodology has grown ever more promising and is now capable of emulating the interplay of urban systems and transportation. Here, we explore the importance of this approach, review how transportation has been factored into or omitted from agent-based models of urban areas, and suggest how it might be handled in future applications. Our approach is to take snapshots of different applications and use these to illustrate how transportation is handled in such models.

Keywords: Agent-based modelling; urban systems; urban modelling

Full Reference:
Wise, S. Crooks, A.T. and Batty, M. (2017). Transportation in Agent-Based Urban Modelling, in Namazi-Rad, M., Padgham, L., Perez, P., Nagel, K. and Bazzan, A. (eds), Agent Based Modelling of Urban Systems, Springer, New York, NY, pp. 129-148. (pdf)

Friday, January 06, 2017

ABMUS2017: The 2nd International Workshop on Agent-based modelling of urban systems

Call for Papers: The 2nd International Workshop on Agent-based modelling of urban systems

The ABMUS2017 workshop on Agent-based modelling of urban systems will be held at the AAMAS2017 conference in Sao Paulo, Brazil on 8-9 May 2017. It is the follow-up of ABMUS2016 held in Singapore during AAMAS2016 on the 10th of May 2016.

Researchers and practitioners who use agent-based models and agent systems to understand, explore, and manage cities and urban infrastructure systems are invited to submit papers to ABMUS2017. The overarching theme for the workshop is data for agent-based models. Data is essential for building, calibrating, and validating agent-based city and urban infrastructure models. But which approaches are optimal for what purposes? We invite presentations that describe data collection and data management approaches in agent-based models, as well as the use of data sets and methodologies that can be translated and re-used between researchers, sectors and countries.

Workshop topics include, but are not limited to, the following:
  • Large scale urban simulation applications
  • Agent-based modelling of urban transport, land-use, housing, energy, health, etc.
  • Spatially explicit micro-simulation modelling
  • Simulation of household behaviour and technology adoption
  • Localized population synthesis
  • Multi-scale urban systems (temporal and spatial)
  • Social simulation of demographic transitions
  • Use of mobile technology to validate activity patterns
  • Techniques for integrating independently developed components
  • Agent-based platforms for urban simulation
  • Data structures for simulating urban environments
  • (Multi-)agent systems for decision support in e.g. transport, energy use and air quality
  • Connection of simulation models to social and geographical theory
  • Development of 'master' city datasets for model validation
At the workshop each presenter will be given 10 minutes to introduce their paper and/or case study, followed by 5-10 minutes in which presenters will share their views on the data for agent-based models theme. After three presentations there will be 20-30 minutes of group discussion in which presenters will act as panel members.

Important dates:
  • 7 February 2017: Deadline for paper submissions
  • 2 March 2017: Notification of acceptance following the review process
  • 17 March 2017: Deadline for submitting camera-ready papers (including LaTeX files)
  • 8-9 May 2017: ABMUS workshop at the AAMAS2017 conference in Sao Paulo, Brazil

For details on how to submit please see or for more information please contact:

The organizing committee consists of:

Wednesday, December 21, 2016

A semester with CSS

This last semester I gave both a graduate and undergraduate course in Computational Social Science (CSS). Both courses survey computational approaches such as system dynamics, social network analysis, machine learning, cellular automata, discrete event simulation, agent-based modeling, and microsimulation to study social phenomena with emphasis on complexity theory.

For the undergraduate class, we met twice a week. In the first class of each week I would outline the topic,  discuss a specific modeling approach and give a range of sample applications. While the second class of the week was devoted to hands on model development (I chose to use NetLogo for this), in order to cement what was discussed in the first class (a learning by doing if you like.). For example, one week we discussed cellular automata (CA) modeling, where in the first class, I outlined its evolution, its basic proprieties and applications (e.g. from simple voting models to that of urban growth). In the second class, the students then built a CA model from scratch.  For the graduate class, more emphasis was placed on theory, critiques of modeling and discussion of key texts.

In both classes, all the students needed to carry out a project where they develop a computational model that investigates a social science research question. This exercise is often their first model (especially for the undergraduates) that many students ever create. Below you can see some of this years models.

Thursday, December 01, 2016

International Congress on Agent Computing

Between the 29th and 30th of November, the International Congress on Agent Computing was held at George Mason University. It was organized to celebrate the 20th anniversary of the publication of Growing Artificial Societies by Robert Axtell and Joshua Epstein. The congress brought together a great line up of interdisciplinary keynote speakers: Brian Arthur, Mike Batty, Stuart Kauffman and  David Krakauer and a  panel discussion entitled "Barriers to Progress in Agent Computing—Technical and Social" with Chris Barrett, Steven Kimbrough, Blake LeBaron, Dawn ParkerFlaminio Squazzoni and Leigh Tesfatsion. Along with the keynotes and there panel there were also over 19 posters and 59 presentations which showcased and demonstrated the theme of the congress, that of the:
"explosive growth of agent modeling over the past two decades in the social sciences, in business and government, and related areas, and offer a tour d’horizon of its present state and myriad applications. Looking forward, we will identify challenges and opportunities — Hilbert Problems, if you will — to shape the future of agent-based computational modeling."

Joshua Epstein and Robert Axtell presenting their works.

Josh and Rob each gave really impressive talks entitled “Agent-­based modeling: From Napkins to Nations” and "The Adoption of Agent Computing over Time by Social Scientists as Compared to Game Theory and Experimental/ Behavioral Economics" respectively. Which reflected how agent computing has evolved over the last 20 years with plenty of funny anecdotes along the way including references and critiques of their works such as "masculine gods of their cyberspace creations" and where the field is going.

What really impressed me about the congress was the atmosphere. That of like minded individuals from many different disciplines coming together and discussing agent computing, complexity and modeling more generally.  Some of this can be seen via photos and tweets of the event.

Alison Heppenstall, Nick Malleson and myslef also participated at the congress with a talk entitled "ABM for Simulating Spatial Systems: How are we doing?" which assessed how has agent-based modeling within the geographical sciences advanced over the last 20 years. Below one can read a brief outline of the talk and a movie of presentation.

While great advances in modeling have been made, one of the greatest challenges we face is that of understanding human behavior and how people perceive and behave in physical spaces. Can new sources of data (i.e. “big data”) be used to explore the connections between people and places?   In this paper we will review of the current state of art of modeling geographical systems.  We highlight the challenges and opportunities through a series of examples that new data can be used to better understand and simulate how individuals behave within geographical systems.

Key Words: Agent-based Modeling, Geographical Information Science, Networks, Cities, Geographical Systems.

Heppenstall, A., Crooks A.T. and Malleson, N. (2016), ABM for Simulating Spatial Systems: How are we doing? International Congress on Agent Computing, 29th-30th, November, Fairfax, VA.

The Growth of Geographical  ABM (selected examples).