One size does not fit all: Customizing MCMC methods in ecological models 

February 28, 2020

Former BIDS Data Science Fellows Lauren Ponisio (now an Assistant Professor in the Department of Entomology at UC Riverside) and Daniel Turek (now an Assistant Professor of Statistics at Williams College), along with BIDS Senior Fellow Perry de Valpine (Associate Professor of Environmental Science, Policy, and Management at UC Berkeley) have published a paper about customizing MCMC methods in ecological models using the open source software NIMBLE and JAGS.

Their new article, One size does not fit all: Customizing MCMC methods for hierarchical models using NIMBLE, addresses the challenges of estimation and inference in ecological modeling.  According to Ponisio, "The application of hierarchical statistical models for analyzing complex data has grown rapidly in recent decades -- however, estimation and inference for these models is not simple. A widely used method is Markov chain Monte Carlo (MCMC) in a Bayesian framework, and improved efficiency of MCMC facilitates all aspects of statistical analysis with Bayesian hierarchical models. In this new paper, we identify strategies to improve MCMC performance, focusing on common ecological models of species occurrence and abundance using open source software NIMBLE and JAGS."

One size does not fit all: Customizing MCMC methods for hierarchical models using NIMBLE
February 14, 2020   |   Ecology and Evolution
Lauren C. Ponisio, Perry de Valpine, Nicholas Michaud, Daniel Turek



Featured Fellows

Lauren Ponisio

Environmental Science, Policy, and Management
BIDS Alum - DATA SCIENCE FELLOW

Perry de Valpine

Environmental Science, Policy, and Management; UC Berkeley
Faculty Affiliate

Daniel Turek

Statistics, ESPM
BIDS Alum - Data Science Fellow