Category Archives: Agent-Based Modeling

Theory, methods, and applications of agent-based models.

Cross-site Comparison of Land-Use Decision-Making

The cumulative effects of local land-use and livelihood changes are a global force of environmental and socio-economic change. Land-use changes result from decisions of individual farmers, pastoralists,  and housing consumers and developers (to name a few). Their decisions are influenced by not only local environmental, social, and economic conditions, but also by far-reaching forces such as economic globalization. The choice of a farmer in Brazil to grow soybeans, for example, can be influenced by the consumption of people in China.

Not all land-uses are created equal. Some have minimum impact on the environment, and some offer sustainable livelihoods for local farmers – finding land uses that accomplish both is difficult. Crafting policies to achieve this two-part goal must contend with both local and global considerations.

Location of one of the study sites near Taoyuan, Hunan Province, China.

A study site near Taoyuan, Hunan Province, China.

On January 29th, my colleagues and I published a paper in PLoS ONE, titled “Cross-site comparison of land-use decision-making and its consequences across land systems with a generalized agent-based model” that describes the development and application of an agent-based virtual laboratory for comparing  land-use and livelihood decision-making processes of rural farmers across geographically distant locations and qualitatively different land-use systems. We use this modeling system across multiple study sites to understand the underlying motivations and rationale of land-use and livelihood decisions of our ‘farmer agents’ and the landscape and livelihood changes that result under various environmental, demographic, and economic scenarios.

Since the traditional mode of scientific experimentation is not feasible with real land-use systems – we are talking about people’s land and livelihoods here – we use simulation-based cross-site comparisons to teach us about what drives the choice of particular land uses and livelihood strategies under different conditions. We use the set of study sites as local examples to synthesis more broadly applicable knowledge of which factors are most important in what contexts.

To explore this question, our investigation had to happen at the decision-making level – a task to which agent-based models are well suited. We also needed a modeling framework that was sufficiently general that it could be applied across multiple locations, yet realistic enough that it could be grounded in real-world data. These needs gave rise to an innovative agent-based virtual laboratory approach that provides a powerful tool for model-based experimentation and synthesis.

Such a model synthesis system can generate the kind of high-level knowledge needed to inform regional policies designed to foster sustainable local land uses and livelihood strategies. Cross-site comparisons use each study site as an example of alternative conditions and/or potential future states, which can aid scenario analysis and the exploration of potential adaptive responses to changing conditions. Furthermore, insights gained from the application of the modeling system to one site can improve our understanding of other similar sites, and foster future research and policy efforts that are sensitive to both the global influence on and local realities of land-use and livelihood change.

Click here to see the web story about this article on SESYNC’s blog.

Advertisements

The Maturation of Socio-Environmental ABMs: Older and Wiser?

After the avalanche of work that was November and December – and the unintentional hiatus from blogging that created – I have now come up for air and am ready to get back to it. Whether it is a particular point in my career or the fact that we are on the eve of a new year, I’m not sure,  but I’m feeling reflective and thought a post about the state of socio-environmental ABMs seemed appropriate. So, here some of my reflections as I look around the landscape of socio-environmental ABMs.

China_FarmerThe first thing that jumps out at me is the explosion of ABM papers in the realm of socio-environmental research. ABMs have achieved widespread use to answer an equally varied assortment of questions ranging from climate change adaptation, agricultural change, urban development, water management, ecosystem services, and on and on …. Fortunately, there are also a number of comprehensive and specialized reviews that have recently been published to help wade through all of this research. Two papers I found particularly useful are reviews of decision-making models in ABMs (An, 2012) and spatial ABMs in socio-ecological systems (Filatova et al., 2013).

As the title of this posts suggests, there also seems to be a maturation in the development and application of ABMs in the socio-environmental realm. I see this maturation in many forms, but two trends in particular are striking. First, the sophistication of these models and the algorithms they employ is incredible. Many ABMs are becoming increasingly computationally intensive. For example, a push for more rigorous sensitivity analyses is leading some to explore the entire parameter space of ABMs with spatially explicit sensitivity analysis (Ligmann-Zielinska, 2013).  I have also seen an increased implementation of sophisticated computer science techniques, such as cognition-based learning algorithms (Magliocca et al., 2011) and encoding of beliefs (Sun and Müller, 2013).

Second, ABMs have become more realistic. A major trend in the ABM field is developing and marketing ABMs for use with policy analysis. Increased availability of highly detailed data sources has led to an explosion of these application-specific (i.e. ‘case-based’) ABMs. Accompanying the development of realistic models has been an advancement of participatory modeling techniques to actively engage in model design and validation (e.g., Zellner et al., 2012).

Yet, this is the point at which I must pause and ask, “The ABM field is clearly older, but is it wiser?”

Despite the popularity of the approach, ABMers still face significant hurdles to having their work and publications accepted. An interesting survey and companion paper was done by Waldherr and Wiejermans (2013). It describes some of the common critiques that ABM researchers still face when trying to get their work published. While some of the critiques demonstrate a lack of understanding on the part of reviewers, they also illustrate areas where much work remains for ABMers to rigorously test and describe their models in order to answer the critics’ questions.

Finally, the abundance of ‘case-based’ ABMs has led some to ask whether additional case-based ABMs are contributing to the ultimate goal of building coherent theory about the structure, dynamics, and sustainability of socio-environmental systems? Is one more place-based model really advancing the community’s knowledge? How do we move forward with models that are both empirically-grounded and general enough to produce theoretical insights? How do we get to these ‘mid-level’ models? These questions, among others, will be the topic of a panel discussion at the upcoming annual meeting of the Association of American Geographers in Tampa, FL, USA from April 8-12, 2014. I have the privilege of accompanying Steve Manson, Tom Evans, David O’Sullivan, Andrew Crooks, Moira Zellner, Li An, and Sarah Metcalf on the panel – looking forward to the discussion!

The promise of the agent-based approach remains high. Applications of ABMs in the socio-environmental context have matured significantly, but we still have a lot of work to do. I am looking forward to the next phase of ABM development in which multiple approaches, algorithms, and techniques are integrated to advance the reliability and usefulness of our models.

New Funding Opportunity: Data-Intensive Analysis and Modeling for Socio-Environmental Synthesis at SESYNC

The National Socio-Environmental Synthesis Center (SESYNC) is inviting proposal submissions for a special funding opportunity designed to support projects pushing the boundaries of computational research in socio-environmental systems. Relevant projects could include (but are not limited to) harmonizing large and/or heterogeneous social and environmental data to answer novel research questions, or developing modeling approaches or applications that are computationally challenging. SESYNC can provide technical support in-house or fund a project team member with sufficient technical skills.

This is an excellent opportunity to push the computational frontiers of your research!

The opportunity listing can be found here: www.sesync.org/opportunities/data-modeling-ses

Submission instructions can be found here: www.sesync.org/opportunities/data-modeling-ses#instructions

NRC Report on Land Change Modeling

Essential reading for all you land change modelers out there!

The report Advancing Land Change Modeling: Opportunities and Research Requirements was released recently in pre-publication format via the National Academies Press web site: http://www.nap.edu/catalog.php?record_id=18385 Additional report info can be found here as well: http://dels.nas.edu/Report/report/18385. The study committee included several geographers, assessed the current state of land-change modeling, and identified opportunities for future developments in these models.

Urban development, agriculture, and energy production are just a few of the ways that human activities are continually changing and reshaping the Earth’s surface. Land-change models (LCMs) are important tools for understanding and managing present and future landscape conditions, from an individual parcel of land in a city to the vast expanses of forests around the world. A recent explosion in the number and types of land observations, model approaches, and computational infrastructure has ushered in a new generation of land change models capable of informing decision making at a greater level of detail. This National Research Council report, produced at the request of the U.S. Geological Survey and NASA, evaluates the various land-change modeling approaches and their applications, and how they might be improved to better assist science, policy, and decision makers.

Exploring Land-Livelihood Transitions

Figure5_rev (2)Rural livelihoods are changing rapidly with economic globalization and global environmental change, which have direct impacts to environmental and socio-economic suitability. All too often the most vulnerable communities – those with the least resources – face the greatest transitions triggered by changing local and global conditions. Those communities also have livelihoods tied to the land, which may lead to environmental degradation and/or fail to support livelihoods in the future. We must advance our understanding of the causes and consequences of land-livelihood transitions in order to avoid maladapted responses that can lead to a loss of land-livelihood sustainability.

My colleagues and I recently published an article in PLoS ONE that explores these issues with an innovative, generalized agent-based model. Because human decision-making drives land-livelihood transitions, a process-level explanation of adaptive responses is needed to explore the conditions under which land-livelihood transitions emerge. In the short-term, this approach advances the use of agent-based virtual laboratories in sustainability research. In coming generations of this modeling approach, we hope to use model insights to devise effective policy interventions aimed at the decision-making level for supporting sustainability .

ABM and GLOBE Project Sessions at the Global Land Project’s 2014 Open Science Meeting

GLP_OSM2014The Global Land Project will hold its second Open Science Meeting (OSM) in Berlin from March 19-21, 2014. This will be a unique opportunity to hear about cutting-edge land and global environmental change research. A list of sessions was recently released here – check it out and see if anything peaks your interest. In particular, I will be co-chairing three sessions related to ABMs, synthesis, and/or GLOBE:

1. Research Session 0126: “Bridging local to global land change studies with the GLOBE online tool” (co-chaired with Erle Ellis)

2. World Cafe Workshop 0075: “From meta-analysis to modeling: understanding local land change globally” (co-chaired with Jasper van Vliet)

3. Short Training Session 0125: “The GLOBE project: evolving new global workflows for land change science” (co-chaired with Erle Ellis and the GLOBE Team)

I attended the first OSM in 2010 (wow, that long ago?!) held at Arizona State University, and it was a great meeting. Session content was exceptional and the meeting was not too big. I highly recommend getting to Berlin next year if you can!

Homo economicus is (mostly) dead

Source: Economists.com/blogs/freeexchange.

Source: Economists.com/blogs/freeexchange.

Do I detect a change in the winds of mainstream economics?

A recent article in the Economist gives me hope. It suggests that ideas of non-rational, adaptive, and distributed decision-making – which have been topics of research in agent-based modeling, psychology, neuroscience, anthropology, and behavioral economics for some time – are now starting to seep into the consciousness of mainstream economics.

Describing Daniel McFadden’s recent work titled “The New Science of Pleasure“, the article details how concepts from psychology, such as prospect theory, are casting renewed doubt on the validity of mainstream economics’ hallmark theory of consumer choice. Indeed, mainstream economic theory has come under fire recently in the wake of economic recession stemming from “irrational” financial decisions, which many economists failed to predict or reconcile with their models and theories.

In all fairness, many mainstream economists would readily offer that their models are unrealistic in many ways, and are useful for understanding how economic systems tend towards rationale outcomes in the long-term. True enough. What this article argues, however, is that the assumptions that underlie mainstream economic models and theory can also lead to unrealistic worldviews and policy recommendations.  For example, ‘more choice is good’, but sometimes this can lead to sub-optimal (i.e. not rational) choices because the consumer is overwhelmed with options. From the article, “Explicitly modelling the process of making a choice might prompt economists to take a more ambiguous view of an abundance of choices.”

And this line of reasoning leads to agent-based modeling as a potential tool to understand how choices are made: what psychological elements influence decisions, how those psychological influence vary with individual heterogeneity characteristics, and how decisions are enacted into behavior.

A parting shot from the article: “This is undoubtedly messier than standard economics. So is real life.”

What is an agent-based virtual laboratory?

I have recently had several conversations with colleagues and exchanges with reviewers about the exact meaning of the term ‘agent-based virtual laboratory.’ So, it seems like the perfect time to devote a single post entirely to unpacking this idea.

First, I should acknowledge that many researchers have used agent-based models (ABMs) of differing levels of detail or abstraction to explore the influence of particular processes or parameters on model outcomes. This is certainly a valuable exercise, and in this sense, the virtual laboratory approach is not new. However, these virtual lab efforts have been undertaken with site-specific, case-study ABMs – and this is where the distinction lies.

“All virtual labs can use ABMs, but not all ABMs are virtual labs.”

That is a favorite phrase I have used in multiple presentations on this topic. It is meant to make a clear distinction between ABMs as tools, and virtual labs as an approach. The type of virtual labs I use in my research and write about here are designed from the outset to be generalized modeling environments that can be applied across many different settings and locations. The advantage of this is that one can conduct comparative research – forming hypotheses of how and under what conditions certain processes or factors will be important or not. These hypotheses can be generated systematically across sites and then tested against empirical data. More conventional virtual lab efforts use case-specific ABMs and thus cannot be easily applied to different sites.

Figure4_rev (2)Take, for example, the figure to the right, which shows agricultural intensification patterns generated by agents with generalized decision models in response to alternative environmental conditions. With this framework, the influence of environmental suitability on land-use intensity – and its interactions with other processes, such as increased market influence and population pressure – can be experimentally explored across different sites.

With that said, if one desires to understand a particular context and/or predict possible future scenarios, the case-specific ABM approach is the way to go. There will always be a place for such models. But those models are not the best option for generating generalized knowledge and building theory. For that, a more generalized and transparent modeling framework is needed.

This approach is similar to that of “artificial systems research” that my friend and colleague Len Troncale describes. Quoting from one of his blog entries, a virtual lab approach “enable[s] adding or subtracting different sets of systems processes to see how these alterations effect sustainability of the resulting systems.” Thus, the goal of agent-based virtual laboratories is to explore and form testable hypotheses of how certain factors interact with agents’ decision-making processes to produce emergent system outcomes, and to do so across different land systems to build towards general land system theory.

Notes from AAG 2013

AAG2013Last week I attended the 2013 annual meeting of the Association of American Geographers in Los Angeles, CA. In particular, I attended the Land Systems Science Symposium and the Agent-Based and Cellular Automata Model for Geographical Systems sessions. It was great to catch-up with old friends and meet a few new colleagues. Now that the chaos of coming back to work after a week off has passed, I thought this would be a good time to reflect on the happenings of the conference.

Overall, I thought this was a much stronger meeting than last year’s. It was apparent from the Land Systems Science (LSS) Symposium that the field formerly known as “Land Change Science” is beginning to come into its own. The full scope of LSS was on display, as the first two days were dedicated to case studies from different world regions, followed on the last two days by LSS modeling, theory, and applications. I was also impressed by the agent-based modeling (ABM) sessions this year. I sense a real transition in ABM research, as the presentations demonstrated much deeper thinking about the implications of model result, and model building as an art and tool for learning. There also seemed to be a sense that ABM is no longer on the fringe – it’s no longer a new method and the ABM community can now discuss the challenges and weaknesses of ABM more comfortably. It was a sure sign that the ABM method and paradigm are maturing.

Several themes in particular caught my attention over the course of the week:

1. YAAWN – Yet Another Agent-based model … Whatever … Nevermind.

That clever acronym came courtesy of the organizers of the panel on ABMs and land-use change modeling organized by David O’Sullivan and Tom Evans. The general motivation for the title was the observation that the number for place-based, case-specific ABMs has exploded, and as a research community, it is worth asking, “What is the marginal gain from one more case-study ABM?” I, of course, was thrilled to hear such a question, as the drive for more systematic, generalized knowledge motivates my use of agent-based virtual laboratories. The question was posed to the panel, and I particularly liked Dan Brown‘s response. The message was that there will always be a role for case-study ABMs, but it is also necessary to balance the use of empirically detailed models with more abstracted models to build theory. This sentiment was reinforced by Sarah Metcalf, who argued that model hybridity was the next wave of ABM research.

2. Attempts to forge systematic ABM practice and knowledge.

Going along with the general observation of higher quality presentations, the thinking about ABM practice was notably deeper this year. I particularly enjoyed David O’Sullivan‘s presentation of creating a ‘pattern language’ for ABMs and cellular automata. The general idea is to create ‘building block’ models out of generalized processes/structures that facilitate the development of more complicated models. I found this idea analogous to Len Troncale‘s work with isomorphies and systems of systems processes theory. Another important question posed by James Millington was, “When should we use ABMs and how complex do they need to be?” Indeed, this is a fundamental question that should be revisited often.

3. Understanding model outcomes and variability.

Finally, another sign that the ABM field is maturing, there was much discussion about the importance of more thoroughly understanding uncertainty and variability in our models. Chris Bone presented an innovative temporal variant/invariant method for evaluating model performance, which shows much promise for deepening our understanding of the sources of variability within complex systems models.

Agent-Based Models in the Real World

thought_process1A recently published News Feature in Proceedings of the National Academy of Sciences by Robert Frederick, titled Agents of Influence, discusses the advancing state of agent-based models (ABMs) and their growing use to inform business and policy decisions. Businesses are employing ABMs to find new efficiencies in complex supply chains, and research efforts to create million-agent models of the economy may soon offer insight into the dynamics of our financial systems and broader economy.

What I like most about this article is that it illustrates how ABMs and complexity thinking are beginning to make their way out of academics and into the real world. A recent example is how Southwest Airlines used ABMs to find more efficient cargo shipping routes, saving the airline millions of dollars. ABMs as virtual laboratories are getting attention, too. The article describes how these models enable decision-makers to explore the consequences of particular business or policy decisions though a range of possible scenarios.

The message is clear: representing heterogeneous, distributed decision-making creates more realistic models, and is enabling researchers, businesses, and policy-makers to navigate complex systems like never before.

Importantly, Frederick does not shy away from the limitations of such models. What is gained in realism by using ABMs often comes at the cost of having to make numerous simplifying assumptions about human behavior. After all, an ABM is only as good as its description of human decision-making processes, which are notoriously unpredictable.

A great closing quote: “Ultimately, … none of these [ABMs] will offer iron-clad predictions, because they have to make simplifying assumptions about human behavior. The true test will be whether those assumptions, and the resulting outputs of the models, convince policymakers to act on their advice.”