Bayesian short course

Welcome to a short course in Bayesian statistics. This course is being taught in March 2014 at Station d’Ecologie Exprimentale du CNRS Moulis. The goal of the course is to use a literate programming style of analysis to carry out statistical tests commonly encountered in the biological sciences. A major reason for using this style is that if you do it well once then you do not have to re-learn it again.

All examples from this course can be reproduced easily. Of course this course is also about Bayesian statistics and the scope is fairly limited, but focused on doing a few things well. Interested participants will be pointed to resources along the way to further their understanding of the different subjects.

Since the majority of the people taking this course are ecologists, much of this website is borrowed from Marc Kery’s book Introduction to WinBUGS for Ecologist [Kery10]. Here is the book’s website where there is also another more advanced book called Bayesian Population Analysis using WinBUGS. It is probably the most accessible piece of Bayesian literature that I have ever read and the examples are quite useful.

See below about installing what you will need to run the examples in this course.

Course Contents

• Bayesian short course - day 1
  • Reproducible research
  • Getting started with Sweave
  • Learning both R and LaTeX with Sweave
• Bayesian short course - day 2
  • An brief introduction
  • Linear Regression

What you will Need

1. LaTeX - a working version of installed

2. R - a current version of the statistical language R

3. Editor - an editor that compiles Sweave

4. BUGS - Though several software are available we recommend JAGS

5. Install rjags and R2jags - See the BUGS page for more info

   Open an R terminal and type the following.

   > install.packages("rjags", dependencies=TRUE)
   > install.packages("R2jags", dependencies=TRUE)
   

Note

As an editor it will be easiest to use RStudio. However, lpEdit may be used for those looking to use languages other than R. It is also possible for those who are inclined to avoid graphical interfaces entirely. Something like Emacs or another such editor could be used so that the examples are run from the command line.

Other helpful materials

• The command line crash course - a book
• Bayesian statistics made (as) simple (as possible) - a youtube video