Bayesian Methods

Catalog description

Bayesian inference provides powerful tools to model random variables. While Bayesian methods often yield the most accurate theoretical results, historically analytical complexity made it challenging to apply Bayesian methods against less trivial problems. Now, the confluence of more powerful computing resources and improved computational algorithms make Bayesian methods the best choice for tackling some of the most complex data science problems. Bayesian analysis is increasingly important in academic research, and research and is the preferred standard statistical analysis tool in data science practice. In this course, students will build from Bayes probability foundations to first applying Bayesian methods to infer binomial probabilities, then hierarchical models, and finally generalized linear models. Students will provide comparisons between frequentist approaches and Bayesian approaches and will build basic algorithms from scratch, as well as using high-performance Markov Chain Monte Carlo (MCMC) methods. Prerequisite(s): Data Acquisition and Management; Computational Statistics and Probability.