The course is intended for students interested in the application of Bayesian methods of data analysis to biological problems. It will provide an introduction to the philosophy behind the Bayesian approach and the techniques used in Bayesian inference. The models and methods of fitting those models to data which underlie the recent emergence of Bayesian inference will be explained, and several topics where Bayesian approaches are central to the development of solutions to biological problems will be covered in more detail. The course will focus on giving the students an understanding of the ideas behind Bayesian approaches, rather than on the technicalities of fitting Bayesian models. The mathematical content will be kept to the simplest level needed. It will be assumed that students have studied simple statistics and data analysis, for example regression and ANOVA.
Students will gain one credit for attending the course. For more details, contact Bob O'Hara (<bob.ohara@Helsinki.FI>, Tel: 050 599 0540).
Dates: 10 lectures, starting February 3th 2003
Times: Monday, 10.00-12.00
Place: Biokeskus III, Room 6201.
Lecture Notes (.pdf files).
I will try and finish removing the typos from the other lectures this week, and then post them here as and when I do them. I'm still surprised that people want to read what I wrote.
February 3rd: Lecture 1, What is probability?
February 10th: Lecture 2, Introduction to Bayesian Inference
February 17th: Lecture 3, More Inference, More Parameters, More Maths!
February 24th: Lecture 4, Model Fitting: MCMC etc.
March 3rd: Lecture 5: Hierarchical Models
March 10th: Lecture 6: Experimental Design and Missing Data
March 24th: Lecture 7: Prediction and Missing Data. (the .pdf seems to be broken at the moment. I'll try and fix it...
March 31st: Lecture 8: Example I: Mark-recapture
April 14th: TWO LECTURES!!!
10am: Lecture 9, Coalesence/phylogeny
WinBugs A popular programme for Bayesian analyses. It uses MCMC to estimate the posterior distribution. The analysis is written in a simple code, so that the computational complexities are hidden from the user. It also has a host of useful examples, which can often be adapted to your own problems.
First Bayes A teaching programme, used to introduce Bayesian ideas.
Hickory, a package for Bayesian estimation of population structure. There are also some notes about the background to F statistisc and the Bayesian approach to their estimation, which look quite nice.