BAYESIAN DATA ANALYSIS
ABOUT THE COURSE
The purpose of this course is to help each student develop a state-of-the-art method for analyzing her own data set, while placing such methods in a Bayesian context. The data examples used in the course are driven by the students and are different every time it is taught, but the Bayesian formalism remains the same. The emphasis is on software rather than theory, especially on Monte Carlo software such as JAGS, and Stan. These free, open-source languages enable one to solve complex data analysis problems without being an expert in either coding or calculus.
The main task of each student will be to learn a statistical package that is appropriate for her problem then apply it to her data. For example, a student who knows Matlab might explore its statistics toolbox and learn to use MatlabStan, a student who prefers ‘R’ might learn to use RJAGS or RStan, and a student who likes Python would learn with PyStan. Students are expected to have some experience in one of Matlab, ‘R’ or Python, but prior experience with Monte Carlo or the packages mentioned above is not required. Knowledge of any particular field of science isn’t required either — the business of fitting data to models spans all areas of science, and knowing the statistical models used in other fields opens a world of possibilities. As statistics has its own jargon, with many synonyms and alternative notations, one of the goals of the course is to increase the student’s vocabulary so that the modern statistical literature is more accessible.