BAYESIAN MODELING FOR WILDLIFE AND FISHERIES

FORS 8984

Fall Semester 2007

Instructor:

Michael J. Conroy
3-427 Forest Resources
542-1167

Lecture:

14:00-16:45 R 1-307 Forestry

Course Objectives

Syllabus

Schedule

9 August -Introductory concepts

16 August -Continued

• Spreadsheet example for beta-binomial

23 August

• Spreadsheet example binomial- logit transformation
• Spreadsheet -Simulating joint posterior for normal
• Berger (1980)- Likelihood principle

30 August --- Priors, posterior simulation
• Binomial simulation in Excel
• Rejection sampling in Python
• Metropolis-Hastings demo in Excel
• Metropolis-Hastings demo in Python

6 September Posterior simulation, MCMC-
• Metropolis-Hastings demo in Excel
• Metropolis-Hastings demo in Python

• Closed CMR M-H in Python

13 September Building basic models in WINBUGS
• Binomial, beta prior
• Poisson, gamma prior
• simulating gamma prior
• simulating normal data
• simulating normal priors
• normal likelihood, diffuse priors
• Nesting data example

  Other

• Coda index file for binomial
• Coda outpu file for binomial
• Binomial prediction/ GOF
• CMR full conditionals

20 September Student Demos with WinBugs

Raccoon Example

Powerpoint for Occupancy Model in BUGS

Dataset for Raccoon Occupancy Model in BUGS

Code for Raccoon Occupancy Model in BUGS

Raccoon script (Clint) Demo for script files in WINBUGS

Double observer point count example

27 September NO CLASS

4 October-- Model convergence diagnostics & goodness of fit

Scab treatment files

binomial GOF

binomial GOF-- multiple samples

Cross validation example

11 October-- Hierarchical modeling

Binomial-beta random effects

Binomial-logit normal random effects

linear trend plus logit normal random effects

multiple samples CMR

PyMC

18 October-- Model selection

binomial model example

Notes on DIC and other model selection criteria

Notes on RJMCMC

see also Wikipedia entry on DIC. Fairly decent quick coverage, points you to the primary literature if you want more.

25 October NO CLASS

1 November -- Model selection; predicting missing values

Simple random sampling w/ missing values

Double random sampling w/ missing response values

Double random sampling w/ missing predictor values

Double random sampling with detection sampling and abundance prediction

King and Brooks - Model selection using RJ-MCMC

Lunn- Generic RJMCMC using graphical models

Reversible jump in WINBUGS

8 November -- student demos

15 November

1400-1445 Intro to Python and PyMC

Overview of Python

Code for examples

Snorkel data

Python/PyMC course at Otago

1445-1530 Felege, Sanderlin, Pacifici model

---- student demos

29 November

1400-1445 Python/ PymC Continued State Space modeling

Binomial model

Binomial simulation model

Binomial random effects model

Regresssion, missing y

Regresssion, missing x

State-space model

1445-1530 Jim Kellner model
6 December
-- PyMC-- State Space modeling

James Martin model

Statistical distributions

Software for Bayes

References

Diversions

Sandbox

• View
• Manage

  ---

  

  Last updated 28 November 2007