Talk by Dr. Andy Hoegh (Mathematical Sciences, MSU)

11/7/2019  4:10-5:00pm  Wilson Hall 1-134


Collective animal movement is characterized by animals that interact and influence each other's movement by attractive and/or repulsive forces. These attractive or repulsive forces can lead to complicated non-linear interaction between individuals that present challenges in fitting traditional statistical models. Agent-based models are defined by a set of agents, in this case animals, that interact based on a pre-defined rules. Viewing agent-based models through the lens of Bayesian state-space models, enables model fitting using particle Markov Chain Monte Carlo (P-MCMC) or Approximate Bayesian Computation (ABC). This talk will provide an overview of statistical agent-based models and discuss computational challenges in estimating parameters.