FISH 506A (3 credits; Section 001) - updated Sept 8, 2010
Course Organizers: Dr Murdoch McAllister, Fisheries Centre, UBC
Schedule: Changed from Term 1 to Term 2, schedule to be determined
Place: Room 320, Aquatic Ecosystems Research Laboratory (AERL), 2202 Main Mall
Course Description:
This course provides an introduction to Bayesian data analysis and statistical modeling methods that are commonly utilized in fisheries stock assessment. Methods covered include approaches that have been applied in fisheries stock assessment to formulate priors, grid-based, importance sampling, and Markov Chain Monte Carlo Methods for integration of posterior distributions for fisheries model parameters, introduction to WinBUGS software for fisheries modeling, diagnostics to assess convergence and goodness of fit, methods to compute Bayes' posteriors (or factors) for alternative fisheries models, fisheries hierarchical models, and Bayesian mark-recapture methods and state-space population dynamics models for fish stock assessment. Minimal entry requirement: first year undergraduate calculus and FISH 504.
Required textbook: Bayesian Methods for Ecology 2007 by M.A. McCarthy
Assessment:
25% Comparing frequentist and Bayesian regression analysis
25% Reparameterizing models to facilitate Bayesian parameter estimation
25% Bayesian hierarchical modeling
25% Bayesian mark-recapture modeling
Topics:
Lecture 1: Introduction to Course and Bayesian and Frequentist Notions of Probability
Practical 1.1: Frequentist and Bayesian estimation of Catch on a Fishing Vessel
Practical 1.2 Bayesian versus Frequentist concepts of probability
Lecture 2: Review of probability theorems and concepts
Practical 2: Exploring the influence of prior density functions on Posterior Density function for Abundance
Lecture 3: Reviewing probability density functions and prior probabilities
Lecture 4: Probability models for data
Lecture 5: Overview of a few "simple" methods to calculate and integrate posterior distributions
Practical 3: Bayesian linear regression analysis in Excel
Lecture 6: Overview of Markov Chain Monte Carlo Methods to calculate and integrate posterior distributions
Practical 4: Introduction to WinBUGS
Lecture 7: More for the Bayesian modeling toolkit: Reparameterizing models and transformation of variables
Practical 5: Reparameterizing models, transforming variables, imputing missing records and using script files in WinBUGS
Lecture 8: Derivation of density functions of transformed variables
Lecture 9: MCMC Modeling Diagnostics
Practical 6: Modeling diagnostics in WinBUGS
Lecture 10: Approaches to model selection and evaluating model plausibility
Practical 7: Bayesian model selection and model uncertainty evaluation methods in WinBUGS
Lecture 11: Introducing the SIR algorithm
Practical 8: Introduction to SIR and comparison with conjugate prior and a simple Hastings-Metropolis Algorithm in Excel
Lecture 12: Review of SIR and Markov Chain, and posteriors for models
Lecture 13: Introduction to Bayesian state-space modeling: illustration using the logistic population dynamics model
Practical 9: Bayesian state-space modelling: illustration using the logistic population dynamics model
Lecture 14: Hierarchical Bayesian methods
Practical 10: Hierarchical modeling exercise on log dorsal shark fin length to log body length
Lecture 15: Bayesian mark-recapture estimation methods and course evaluation