AMS 206 (Bayesian Statistics) at UC Santa Cruz, Winter 2005

DEPARTMENT OF
APPLIED MATHEMATICS & STATISTICS (AMS)

AMS 206: Bayesian Statistics (Winter 2005)

A graduate-level introduction to Bayesian statistical methods for inference and prediction. Exchangeability. Prior, likelihood, posterior, and predictive distributions. Coherence and calibration. Conjugate analysis. Markov Chain Monte Carlo methods for simulation-based computation. Hierarchical modeling. Bayesian model diagnostics, model selection, and sensitivity analysis.

Instructor: David Draper (Baskin Engineering 135, 459-1295, draper@ams.ucsc.edu)
TA: Milovan Krnjajic (Baskin Engineering 137, 459-1308, milovan@ams.ucsc.edu)
Lecture/discussion meetings: MWF 2:00-3:10pm, room 192 of the Engineering 2 Building, plus some extra meetings (to be arranged) to make sure we cover everything that needs to be covered (I will need to be out of town on research business during several of our usual MWF slots).
Office hours:
- Instructor: Wed 9-10am, Thu 9-10am in JB 135
- TA: Tue 12-1pm, Fri 12:30-1:30pm in JB 137
News and supplemental reading:
- (21 Jan) For people who would like to read ahead in the class, here are two handouts we'll cover later (this is optional reading, for now):
  - Bayesian Hierarchical Modeling (chapters 1 and 2 from a book I'm working on)
  - MCMC in Practice (chapter 1 from a book on Markov chain Monte Carlo by Gilks et al.)
- (25 Jan) (more optional reading): If you'd like to see more about the use of Bayesian decision theory and maximization of expected utility in a complicated real-world problem, the following paper (which also contains a review of three of the leading stochastic optimization methods) may be of interest to you:
  - Fouskakis D, Draper D (2002). Stochastic optimization: a review. International Statistical Review, 70, 315-349.
- (4 Feb) (more optional reading): Here's a paper I mentioned in class recently, on the Bayesian approach to acknowledging uncertainty in the statistical model itself:
  - Draper D (1995). Assessment and propagation of model uncertainty (with discussion). Journal of the Royal Statistical Society Series B, 57, 45-97.
- (8 Mar) The latest version of WinBUGS (1.4.1) has been installed on the following machines in the School of Engineering graduate student computing lab in JB 340, for your use in this class: Congo, Cypress, Donau, Maple, and Walnut.
- (8 Mar) (more optional reading) If you'd like to delve more deeply into DIC, here's the discussion paper on which it's based:
  - Spiegelhalter DJ, Best NG, Carlin BP, van der Linde A (2002). Bayesian measures of model complexity and fit (with discussion). Journal of the Royal Statistical Society, Series B, 64, 583-639.
- (8 Mar) Here's both parts of the CODA manual:
  - Best N, Cowles MK, Vines K (1996). CODA: Convergence Diagnostics and Output Analysis Software for Gibbs Sampling Output (version 0.30). MRC Biostatistics Unit, University of Cambridge (UK).
  - Best N, Cowles MK, Vines K (1997). CODA: Convergence Diagnostics and Output Analysis Software for Gibbs Sampling Output (version 0.40, addendum to manual). MRC Biostatistics Unit, University of Cambridge (UK).
Course handouts: (in PDF format; please let me know if you have any problems with the resolution of the downloaded files, or any other difficulties)
- Meeting 1 (Mon 10 Jan 2005):
  - Handout 1 (Course introduction, week-by-week plan, reading list)
  - Handout 2 (Participant background survey, part 1)
  - Handout 3 (Participant survey, part 2)
  - Handout 4 (Printed lecture notes, part 1: Background and Basics)
  - Handwritten notes 1 (Supplemental class notes from 10 Jan)
- Meeting 2 (Wed 12 Jan 2005):
  - Handout 5 (Homework assignment 1)
  - Handwritten notes 2 (Supplemental class notes from 12 Jan)
- Meeting 3 (Fri 14 Jan 2005):
  - Handout 6 (Chapter 9, "Expected Value, Standard Error, and the Central Limit Theorem", from Draper D (in preparation), Thinking About Uncertainty: An Introduction to Probability and Statistics)
  - Handwritten notes 3 (Supplemental class notes from 14 Jan)
- Meeting 4 (Wed 19 Jan 2005):
  - Handout 7 (Bayesian spam filtering)
  - Handout 8 (Printed lecture notes, part 2, installment 1 (pages 1-20))
  - Handwritten notes 4 (Supplemental class notes from 19 Jan)
- Meeting 5 (Fri 21 Jan 2005):
  - Handout 9 (Printed lecture notes, part 2, installment 2 (pages 21-40))
  - Handwritten notes 5 (Supplemental class notes from 21 Jan)
- Meeting 6 (Mon 24 Jan 2005):
  - Handout 10 (Page 1 of Draper D (in preparation), Bayesian Hierarchical Modeling (please see the news item on 21 Jan above to download the full document))
  - Handout 11 (Homework assignment 2)
  - Handout 12 (Printed lecture notes, part 2, installment 3 (pages 41-60))
  - Handwritten notes 6 (Supplemental class notes from 24 Jan)
- Meeting 7 (Wed 26 Jan 2005):
  - Handout 13 (Appendix on common distributions from Gelman et al.)
  - Handout 14 (Judgment Under Uncertainty: Heuristics and Biases)
  - Handout 15 (Adding Art to the Rigor of Statistical Science)
  - Handwritten notes 7 (Supplemental class notes from 26 Jan)
- Meeting 8 (Fri 28 Jan 2005):
  - Handout 16 (Subconsciously, Athletes May Play Like Statisticians)
  - Handout 17 (Schedule for AMS 206, winter 2005)
  - Handout 18 (Solutions to homework 1 (not available on the web))
  - Handwritten notes 8 (Supplemental class notes from 28 Jan)
- Meeting 9 (Mon 31 Jan 2005):
  - Handout 19 (Printed lecture notes, part 2, installment 4 (pages 61-80))
  - Handwritten notes 9 (Supplemental class notes from 31 Jan)
- Meeting 10 (Wed 2 Feb 2005):
  - Handout 20 (Printed lecture notes, part 2, installment 5 (pages 81-100))
  - Handwritten notes 10 (Supplemental class notes from 2 Feb)
- Meeting 11 (Thu 3 Feb 2005):
  - Handwritten notes 11 (Supplemental class notes from 3 Feb)
- Meeting 12 (Fri 4 Feb 2005):
  - Handwritten notes 12 (Supplemental class notes from 4 Feb)
- Meeting 13 (Mon 7 Feb 2005 (morning)):
  - Handout 21 (Printed lecture notes, part 2, installment 6 (pages 101-117))
  - Handout 22 (Homework assignment 3)
  - Handout 23 (How to plot multiple functions on the same graph in R)
  - Handwritten notes 13 (Supplemental class notes from 7 Feb (morning))
- Meeting 14 (Mon 7 Feb 2005 (afternoon)):
  - Handout 24 (Printed lecture notes, part 3, installment 1 (pages 1-20))
  - Handwritten notes 14 (Supplemental class notes from 7 Feb (afternoon))
- Meeting 15 (Wed 9 Feb 2005)
  - Handout 25 (Mid-course survey of form and content)
  - Handout 26 (Printed lecture notes, part 3, installment 2 (pages 21-40))
  - Handwritten notes 15 (Supplemental class notes from 9 Feb)
- Meeting 16 (Fri 11 Feb 2005)
  - Handwritten notes 16 (Supplemental class notes from 11 Feb)
- Meeting 17 (Mon 14 Feb 2005)
- Meeting 18 (Wed 16 Feb 2005)
  - Handout 29 (RSS News editorial: Peace of mind)
  - Handout 30 (Chapters from Gilks et al. on MCMC)
  - Handwritten notes 18 (Supplemental class notes from 16 Feb)
- Meeting 19 (Fri 18 Feb 2005)
  - Handwritten notes 19 (Supplemental class notes from 18 Feb)
- Meeting 20 (Wed 23 Feb 2005)
  - Handout 31 (Printed lecture notes, part 3, installment 4 (pages 61-87))
  - Handwritten notes 20 (Supplemental class notes from 23 Feb)
- Meeting 21 (Thu 24 Feb 2005)
  - Handwritten notes 21 (Supplemental class notes from 24 Feb)
- Meeting 22 (Fri 25 Feb 2005)
  - Handout 32 (Printed lecture notes, part 4, installment 1 (pages 1-20))
  - Handout 33 (Science-based prediction for complex systems: article from Los Alamos Science)
  - Handwritten notes 22 (Supplemental class notes from 25 Feb)
- Meeting 23 (Mon 28 Feb 2005)
  - Handout 34 (Homework assignment 4)
  - Handout 35 (part 1) (BUGS 0.5 manual (version ii))
  - Handout 35 (part 2) (BUGS 0.6 addendum to manual)
- Meeting 24 (Wed 2 Mar 2005)
  - Handout 36 (Solutions to homework 3 (not available on the web))
  - Handwritten notes 23 (Supplemental class notes from 2 Mar)
- Meeting 25 (Fri 4 Mar 2005)
  - Handout 37 (Printed lecture notes, part 4, installment 2 (pages 21-40))
  - Handwritten notes 24 (Supplemental class notes from 4 Mar)
- Meeting 26 (Mon 7 Mar 2005 (morning))
  - Handwritten notes 25 (Supplemental class notes from 7 Mar (morning))
- Meeting 27 (Mon 7 Mar 2005 (afternoon))
  - Handout 38 (Page 1 of the CODA manual (please see the news item on 8 Mar above to download the full document))
  - Handout 39 (Printed lecture notes, part 4, installment 3 (pages 41-60))
  - Handout 40 (Default project)
  - Handout 41 (Book review from the New Yorker: Select all)
  - Handwritten notes 26 (Supplemental class notes from 7 Mar (afternoon))
- Meeting 28 (Wed 9 Mar 2005)
  - Handout 42 (Printed lecture notes, part 4, installment 4 (pages 61-72))
  - Handout 43 (Printed lecture notes, part 5)
  - Handwritten notes 27 (Supplemental class notes from 9 Mar)
- Meeting 29 (Fri 11 Mar 2005)
  - Handout 44 (Download from the New York Times: Looking Inside the Brains of the Stingy)
  - Handout 45 (Discussion by Brian Dennis from Ecological Applications: Should Ecologists Become Bayesians?)
  - Handout 46 (Article by Jim Clark from Ecology Letters: Why Environmental Scientists Are Becoming Bayesians)
  - Handwritten notes 28 (Supplemental class notes from 11 Mar)
- Meeting 30 (Mon 14 Mar 2005)
Data sets and WinBUGS code:
- NB10 case study
- In-home geriatric assessment (IHGA) case study
- Teacher expectancy case study
- Data from Berkeley traffic case study for default project
  - traffic.txt

Last update: 15 March 2005