140.763 - Third Term 2002
BAYESIAN METHODS
Instructor: Francesca Dominici
Teaching Assistant: Matilde Trevisani
LECTURE: W4007 Tu-Th 1:30 - 3:00 pm
LAB: Th W2033 4:00 - 5:00 pm
Please send to Matilde (matildet@stat.unipd.it) your e-mail address to be included in the class list
INDEX
Course info
Announcements
Lecture Notes
Lab Notes
Data sets
Software (S+,R,BUGS)
Bayesian Methods books
COURSE INFO
  • SYLLABUS [ps] [pdf]
  • Dr. Dominici's Office Hour Th 3:00 - 4:00 pm
  • Dr. Trevisani's Office Hour Tu 4:00 - 5:00 pm
  • ANNOUNCEMENTS
  • Final will be posted here TUESDAY 3/5 [ps] [pdf]
  • Course Evaluations will be distributed on TUESDAY 3/5 Please come to the class and give me feedback
  • Deadline for returning the final is TUESDAY 3/19, please leave your exam in my mailbox

    LECTURE NOTES
  • 1. Setting up a probability model, Bayes'rule, posterior means and variances, binomial model [ps] [pdf]
  • HW1 [ps] [pdf]
  • Solutions of HW1 [ps] [pdf]
  • 2. Standard univariate models including the normal model, conjugate and noninformative prior distribution [ps] [pdf]
  • 3. Multiparameters models, normal with unknown mean and variance, the multivariate normal distribution, multinomial models. [ps] [pdf]
  • HW2 [ps] [pdf]
  • Solutions of HW2 [ps] [pdf]
  • 4. Hierarchical models, estimating populations parameters from the data [ps] [pdf]
  • MID-TERM ASSIGNMENT [ps] [pdf]
  • MID-TERM SOLUTIONS ( Courtesy of Hongfei Guo ) [ps] [pdf]
  • 5. Posterior simulation and integration [ps] [pdf]
  • 6. Markov Chain Simulation [ps] [pdf]
  • READING ASSIGNEMENTS
  • Sampling-Based Approaches to Calculating Marginal Densities by Gelfand A. and Smith A.F.M. JASA 1990 [pdf]
  • Explaining the Gibbs Sampler by Casella G. and George E.I. The American Statistician 1992, vol 46, pp:167-174
  • Understanding the Metropolis Algorithm by Chib S. and Greenberg E. The American Statistician 1995, vol 49, pp:327-335
  • 7. Re-analyses of the data sets used in the papers below by implementing Gibbs Sampling and Metropolis Algorithm [ps] [pdf]
  • Illustration of Bayesian Inference in Normal Data Models Using Gibbs Sampling by Gelfand A.E., Hills S.E., Racine-Poon A., Smith A.F.M., JASA, Vol. 85, pp. 972-985. [ps]
  • A Generalization of the Probit and Logit Methods for Dose Response Curves Prentice P., Biometrics, Vol. 32, pp. 761-768. [ps]
  • 8. Bayesian linear regression analysis, hierarchical linear regression models, Bayesian variable selection [ps] [pdf]
  • Variable Selection Via Gibbs Sampling George E.I. and McCulloch R.E. JASA Vol.88 pp. 881-889 [ps]
  • 9. Generalized linear models: hierarchical logistic regression, hierarchical log-linear regression
  • Bayesian Analyses of the rat tumor data and of air pollution and mortality national data base [ps] [pdf]
  • 10. Bayesian Semi-Parametric Logistic Regression [ps] [pdf]
  • 11. Study design, missing data [ps] [pdf]
  • 12. Review [ps] [pdf]

    LAB NOTES
    WEEK 1
  • Introduction to R (R 1.4.0 for Windows, R 1.3.1 for Sparc stations) and Bayesian Analysis of Binomial Data in R [lab1.pdf] [slides1.pdf] [slides14.pdf] [placenta.r] [betabin.r]
    WEEK 2
  • Introduction to BUGS (WinBUGS 1.3 for Windows, Classic BUGS 0.603 for Sparc stations). One-parameter models: 1) normal with known variance (R and BUGS); and 2) Poisson (BUGS). [lab2.pdf] [slides2.pdf] [norm1.b] [pois.b]
    WEEK 3
  • Multi-parameter models: 1) normal with unknown mean and variance (R and BUGS); 2) multinomial model (R and BUGS) [lab3.pdf] [slides3.pdf] [norm2.r] [norm2.dat] [norm2.b] [mnomial.r] [mnomial.b]
    WEEK 4
  • Hierarchical priors for pooling strength: a meta-analysis example (R and BUGS) [lab4.pdf] [slides4.pdf] [N-N.r] [N-N.b]
    WEEK 5
  • Hierarchical linear models: a growth curve analysis using the Gibbs sampler (R and BUGS) [lab5.pdf] [slides5.pdf] [rats.dat] [Rats: a Normal hierarchical model] [Birats: a bivariate Normal hierarchical model]
    WEEK 6
  • Metropolis algorithm for generalized nonlinear models: the generalized logit model in BUGS [lab6.pdf] [slides6.pdf] [Beetles: logistic, probit and extreme value (log-log) model comparison] [beetles.b]
    WEEK 7
  • Miscellanea in BUGS: a probit model via latent variable and stochastic search variable selection via a hierarchical normal mixture model [lab7.pdf] [slides7.pdf] [latentv.b] [SSVSsim.b]
    WEEK 8
  • Multilevel GLM for hierarchical data sets [lab8.pdf] [slides8.pdf] [Program8.1] [MLpois.b]

    DATA SETS FROM TEXT BOOK
  • Football scores and point spreads (Figure 1) [football.data]
  • Speed of Light measurements (Table 3.1) [light.data]
  • Rat Tumors (Table 5.1) [rat.data]
  • Clinical Trials of beta-blockers (Table 5.4) [betablockers.data]
  • Baseball batting (Table 6.1) [baseball.data]
  • Congressional Elections and incumbency (Section 8.4) [election.data]
  • Forecasting Presidential Elections Section 13.2) [forecast.data]
  • Contingency Table from a Sample Survey (Table 14.2) [cont_table.data]
  • Cities and Town in New York State (Section 18.3) [newyork.data]
  • Rats: a normal hierarchical model (BUGS Examples, Volume I and II) [rats.dat]
  • Beetles data set [beetles.dat]
  • Finney's vasoconstriction data [vasoconstriction.dat]
  • SOFTWARE (S-PLUS)
  • Posterior inferences under a Binomial model [placenta.S]
  • Posterior inferences under a Poisson model [poisson.S]
  • Posterior inferences under a Normal model [normalnormal.S]
  • Sample from a Multivariate Normal Distribution [multnorm.s]
  • Sample from a Wishart and Inverse Wishart Distributions [Wishart.s]
  • Sample from a Dirichlet Distribution [Dirichlet.s]
  • Posterior inferences under a Normal model [normalnormal.S]
  • Bayesian Analysis of a Biossay Experiment [biossay.S] [commands.biossay.S]
  • Estimating the risk of tumor in a group of rats [tarone.S]
  • Hierarchical normal model with unknown variance: analysis of the diet measurements with a Gibbs Sampling [hierarnorm.gibbs.S]
  • Bayesian Linear Regression Analysis of Radon Data [radon.S]
  • Implement Importance Sampling [importance.S]
  • Approximating the Posterior Distribution of all Unknown Parameters under a Hierarchical Logistic Model: Estimating the risk of tumor in a group of rats [hlogistic.S]
  • Implement Metropolis [metropolis.S]
  • Implement a Gibbs Sampling [babymcmc.S]
  • Implement Gibbs Sampling under Bivariate Normal Model [Gibbs.S]
  • TEXT BOOKS
  • Bayesian Data Analysis Gelman, Carlin J., Stern and Rubin (TEXTBOOK) [table of contents]
  • Markov Chain Monte Carlo in Practice, Gilks, Richardson, and Spiegelhalter [table of contents]
  • Bayes and Empirical Bayes Methods for Data Analysis, Carlin B. and Louis T. [table of contents] [description]
  • Modeling in Medical Decision Making: A Bayesian Approach, Parmigiani G. [table of contents]
  • Bayesian Statistical Modelling, Congdon P. [table of contents]