An agent-based framework that aggregates social network-level individual interactions to run targeting and rewarding programs for a freemium social app. Git source code in https://bitbucket.org/mchserrano/socialdynamicsfreemiumapps

The purpose of this model is to investigate mechanisms driving the geography of educational inequality and the consequences of these mechanisms for individuals with varying attributes and mobility.

The model aims at estimating household energy consumption and the related greenhouse gas (GHG) emissions reduction based on the behavior of the individual household under different operationalizations of the Theory of Planned Behaviour (TPB).
The original model is developed as a tool to explore households decisions regarding solar panel investments and cumulative consequences of these individual choices (i.e. diffusion of PVs, regional emissions savings, monetary savings). We extend the model to explore a methodological question regarding an interpretation of qualitative concepts from social science theories, specifically Theory of Planned Behaviour in a formal code of quantitative agent-based models (ABMs). We develop 3 versions of the model: one TPB-based ABM designed by the authors and two alternatives inspired by the TPB-ABM of Schwarz and Ernst (2009) and the TPB-ABM of Rai and Robinson (2015). The model is implemented in NetLogo.

We build a stylized model of a network of business angel investors and start-up entrepreneurs. Decisions are based on trust as a decision making tool under true uncertainty.

The MeReDiem model aims to simulate the effect of socio-agricultural practices of farmers and pastors on the food sustainability and soil fertility of a serrer village, in Senegal. The model is a central part of a companion modeling and exploration approach, described in a paper, currently under review)

The village population is composed of families (kitchens). Kitchens cultivate their land parcels to feed their members, aiming for food security at the family level. On a global level , the village tries to preserve the community fallow land as long as possible.

Kitchens sizes vary depending on the kitchens food production, births and migration when food is insufficient.
…

Micro-targeted vs stochastic political campaigning agent-based model simulation. Written by Toby D. Pilditch (University of Oxford, University College London), in collaboration with Jens K. Madsen (University of Oxford, London School of Economics)

The purpose of the model is to explore the various impacts on voting intention among a population sample, when both stochastic (traditional) and Micto-targeted campaigns (MTCs) are in play. There are several stages of the model: initialization (setup), campaigning (active running protocols) and vote-casting (end of simulation). The campaigning stage consists of update cycles in which “voters” are targeted and “persuaded” - updating their beliefs in the campaign candidate / policies.

This adaptation of the Relative Agreement model of opinion dynamics (Deffuant et al. 2002) extends the Meadows and Cliff (2012) implementation of this model in a manner that explores the effect of the network structure among the agents.

This model implements two types of network diffusion from an initial group of activated nodes. In complex contagion, a node is activated if the proportion of neighbour nodes that are already activated exceeds a given threshold. This is intended to represented the spread of health behaviours. In simple contagion, an activated node has a given probability of activating its inactive neighbours and re-tests each time step until all of the neighbours are activated. This is intended to represent information spread.

A range of networks are included with the model from secondary school friendship networks. The proportion of nodes initially activated and the method of selecting those nodes are controlled by the user.

PolicySpace models public policies within an empirical, spatial environment using data from 46 metropolitan regions in Brazil. The model contains citizens, markets, residences, municipalities, commuting and a the tax scheme. In the associated publications (book in press and https://arxiv.org/abs/1801.00259) we validate the model and demonstrate an application of the fiscal analysis. Besides providing the basics of the platform, our results indicate the relevance of the rules of taxes transfer for cities’ quality of life.

The wisdom of the crowd refers to the phenomenon in which a group of individuals, each making independent decisions, can collectively arrive at highly accurate solutions—often more accurate than any individual within the group. This principle relies heavily on independence: if individual opinions are unbiased and uncorrelated, their errors tend to cancel out when averaged, reducing overall bias. However, in real-world social networks, individuals are often influenced by their neighbors, introducing correlations between decisions. Such social influence can amplify biases, disrupting the benefits of independent voting. This trade-off between independence and interdependence has striking parallels to ensemble learning methods in machine learning. Bagging (bootstrap aggregating) improves classification performance by combining independently trained weak learners, reducing bias. Boosting, on the other hand, explicitly introduces sequential dependence among learners, where each learner focuses on correcting the errors of its predecessors. This process can reinforce biases present in the data even if it reduces variance. Here, we introduce a new meta-algorithm, casting, which captures this biological and computational trade-off. Casting forms partially connected groups (“castes”) of weak learners that are internally linked through boosting, while the castes themselves remain independent and are aggregated using bagging. This creates a continuum between full independence (i.e., bagging) and full dependence (i.e., boosting). This method allows for the testing of model capabilities across values of the hyperparameter which controls connectedness. We specifically investigate classification tasks, but the method can be used for regression tasks as well. Ultimately, casting can provide insights for how real systems contend with classification problems.