Stat 4860/6860: Introduction to Bayesian Statistical Analysis

Week 10:

• This week we will see how to run a script in WinBUGS for the O-ring data and to compare models using the DIC.
• This week we will consider the One-way Random Effects ANOVA model.  The handouts from the 2005 JSM poster presentation are available at onewayrand
• And we will consider the Bayesian approach to Diagnostics Testing.
• Homework 5 has been posted.
• The class final will be on Monday August 28, 7 to 9pm.

Review for the Final.  Read the following

• Guidance for the Use Bayesian Statistics in Medical Device Clinical Trials
• Simon Jackman, Stanford University, Estimation and Inference via Bayesian Simulation: An Introduction to Markov Chain Monte Carlo, American Journal of Political Science, Vol. 44, No. 2, April 2000, pp. 369-398.  icpsr99.pdf

Week 9:

• This week everyone should settle on a Project to work on.
• Two-sample problems Lee  WinBUGSprograms: sleep.txt  ratdiet.txt  ratdiet_unknownvar.txt
• Has everyone tried out BetaBuster or the following webpages from Epi Tools, BetaParams1 BetaParams2.
• This week we will continue to discuss Linear Regression. regression.ssc RegressionBugs.odc RegressionDoodle.odc
• This week we will discuss Bayesian Logistic Regression.  Isvee3-Logistic-Regr1.ppt
• DIC: Deviance Information Criteria  Consider the pumps2.odc

Week 8:

• Class will be cancelled on Monday August 7.
• The take-home midterm is due Wednesday.

Week 7:

• Project: The project for the class will be to find a topic of interest to you and to find an example WinBUGS program to understand, run, and interpret the results from.  To start you might look over the examples in WinBUGS. Next week on Wednesday a proposal for your project will be collected which should include 1. The topic of interest.  That is the area of application of you project.  2. Give a description of the statistical methods used in the example program you have decided to work on.  3. Describe the data and the parameters in the model.  4.  Describe what the parameters if interest are in the model and what the possible inferences will be.
• Homework 4 has been posted.
• Take a look at this website for information about Diagnostic Testing. BEST  And try BetaBuster for beta prior selection.

Week 6:

• Midterm on Wednesday.
• Homework 3 will be collected next week on Wednesday.

Week 5:

• Some hw solutions have been posted on Blackboard under Course Materials.
• There is a Stats Magazine article posted on Blackboard under Course Materials on the Relationship between x.bar and s.
• There is a new R handout on the normal Bayes problem using the Bolstad library.
• Two new WinBUGS programs have been posted.  BetaBinomialodds.odc pumps.odc

Week 4:

• Homework 2 will be collected on Wednesday.
• This week we began working on class quizzes on developing appropriate models for the likelihood and the discussion of conjugate priors and reference priors.
• The quiz that was given on Monday: quiz1.pdf
• The quiz that was given on Wednesday: quiz2.pdf
• Please see the Homework link for the newly posted homework assignment 3.

Week 3:

• Homework 1 is due today.
• Install the library Bolstad in R.  See the Handouts link for examples using the Bolstad library.
• Today we will go to the computer lab to try R with the Bolstad library (hopefully) and WinBUGS.

Week 2:

• Install R on your computer.  See the Statistics Software link.  I have also posted links to Tinn-R and JGR.
• There is an updated Introduction to R posted under Handouts with individual programs for each part of the handout.  Run all of the programs and make a list of things that are unclear.
• From the r-project website be sure to take a look at the Manuals page.
• Run the program bayes.r to check your answers to the 3 Expercises.  See the Handouts link.
• There are 3 handouts on random number generation posted, see the Handouts link.
• On example where Bayes Theorem is applied these days is in Bayesian Spam Filtering
  • LINUX Journal SpamBayes
  • A Plan for Spam
  • Better Bayesian Filtering
• Here is a link the spam filtering in the Mozilla browser.  It is SpamBayes that they use.  spam
• Install WinBUGS14 on your computer.  See the Statistics Software Link. 
• See the Handouts link for some WinBUGS examples.
• Homework 1 will be collected on Monday July 3.  Well hopefully.

Week 1:

• A reading assignment has been posted on Blackboard under Course Materials.
• See the Homework link for the first assignment.
• Install the necessary software for the class. R, WinBUGS

Week 0:

• The book for Statistics 4860 is Introduction to Bayesian Statistics by William M Bolstad, Wiley, 2004
• Bolstad
• The book for Statistics 6860 is Bayesian Statistics and introduction by Peter Lee, Hodder Arnold, 2004.
• Lee     Rprograms     WinBUGSprograms
• Two other references for the course:
  • Bayesian Statistical Modelling by Peter Congdon, 2001.
  • Bayesian Data Analysis, Second Edition, by Gelman, Carlin, Stern, and Rubin.
    • Andrew Gelman
    • Hal S. Stern