Bayesian Data Analysis (Stat 668)
Spring 2015

General Information

Lecturer Zhiqiang Tan
Office: 459 Hill Center
Email: ztan at

Lectures: MW 1:40-3:00PM, HIL 522
Office hours: TH 1:00-2:00PM.
Prerequisite Stat 593 or equivalent.
Textbook Gelman et al (2014) Bayesian Data Analysis (3rd edition), CRC Press.
Parametric models Chapters 2-3
Hierarchical models Chapter 5
Model checking Chapters 6-7
Bayesian computation Chapters 10-13
Regression models Chapters 14-16
Mixture models Chapter 22
Exams and Homework There will be 2 midterm projects and a final project.
Homework will be assigned and collected.
Grading The final grade will be based on the following components with the weights (corrected March 4):
Homework: 60%
Final Project: 40%
Makeup policy Make-up exams will only be given if written documentation of a major outside circumstance is provided by a dean or a doctor.
Students who miss exams without presenting proper documentation in a timely manner will receive a grade of zero.
Homework Assignments HW 1, HW 2, HW 3, HW 4, HW 5.

Selected solution to HW 1.

Here is Gelman et al. (2008).
Here is Karim & Zeger (1992). The dataset is here.
Here is Booth & Hobert (1999).
Announcements Posted Feb 12:
        R codes for Metropolis sampling (here) and Gibbs sampling (here) from bivariate normal distributions.
Posted Feb 18, corrected March 4:
        R codes for Gibbs sampling (here) for posterior simulation in the eight-school example.
Posted Feb 23:
        R codes for Gibbs sampling (here) and Metropolis sampling (here)for posterior simulation in the coagulation example.
Posted March 4:
        R codes for parameter-expanded Gibbs sampling (here) for posterior simulation in the eight-school example.
Posted March 25, revised Apr 1:
        R codes for the presidential election example (here).
        The datasets are here and here.
Posted Apr 16:
        R codes for logit and probit regression (here and here).
        The dataset is here.
Posted Apr 22:
        R codes for the salamander example (here).
Posted May 7:
        R codes for the eight-school example, with model checking and comparison (here).