Applied Mathematics, University of Colorado-Boulder

Prof. Vanja M. Dukic  

Appm 7400-001    Spring 2012
http://amath.colorado.edu/courses/7400_001/2012Spr/

 

Bayesian Statistics and Computing

 

Professor: Vanja M. Dukic

email: Vanja.Dukic (a t) colorado.edu

Office hours: MW 1:00pm-2:30pm, Applied Math, ECOT 322


This class is an MS/PhD level class, but is intended to be practical and catered to audiences from mixed disciplines. It will cover the basics of Bayesian inference, modeling, and algorithms. It will begin with the introduction to Bayesian statistics, and cover normal and non-normal approximation to likelihood and posteriors, the EM algorithm, data augmentation, and Markov chain Monte Carlo (MCMC) methods. We will also cover some more advanced MCMC and particle filtering algorithms towards the end of the semester. Multiple Bayesian modeling examples (hierarchical linear, hierarchical generalized linear, time series, hidden Markov, and state space models) will be used throughout the course. There will be weekly homeworks, and students will be expected to complete a data analysis project by the end of the course. There will be one final in-class presentation. Variety of programming languages (eg, Matlab, R, C, C++) can be used to implement the algorithms and carry out data analyses. Previous coursework in probability and statistics (statistical modeling), as well programming experience are required.

 

Required text:

"Tools for Statistical Inference ", by Martin Tanner.

Grading policy:    

Homework: 45%

Final presentation and project:   45%

Course Participation:  10%

Final presentations::  20-min talks on Monday (4/30) and Wednesday (5/2) from 4:00-6:30pm. Room ECCR 1B51.

Final papers due by::  Saturday (5/5) at 5pm, via email (pdf files only please).