An Introduction to Modern Statistical Analysis using Bayesian Methods

Inaugural Social Sciences Computing Hub Workshop

This presentation provides an introduction to modern data analysis with emphasis on Bayesian approach to statistical inference. The need for statistical analysis is overwhelming in today’s world given the enormous amounts of data collected in a variety of application areas such as finance, medicine, engineering, social sciences, ecology, marine science, bioinformatics and many more. The goals of data analysis may be manifold and include hypothesis testing, summarizing of the collected data, mathematical modelling of the phenomenon of interest, or prediction of the future data. We shall point out the differences between classical and Bayesian data modelling and analysis, and look at the fusion of the advantages from both approaches. Bayesian analysis incorporates a fully probabilistic framework for quantifying uncertainty in terms of probability distributions combining available prior information with the observed data. Baysian modelling methods rely on simulation-based inference implemented using Monte Carlo Markov Chain (MCMC) algorithms. This requires support of substantial computational power which has been available to applied researchers accross the board, in the last two decades thanks to a rapid growth of information technology.

The presentation consists of two lectures with demonstrations (60 minutes each) inclusive of time for questions and discussion.

Dr. Milovan Krnjajic is SFI Lecturer in Statistics at the School of Mathematics, NUI Galway. He holds a Ph.D. degree in Statistics from the University of California, Santa Cruz. Dr. Krnjajic teaches courses in Probability theory, Stochastic Processes, and Statistics, including a course on Bayesian statistical methods. His research is focused on Bayesian methods for analysis of structured data, in particular on the development of Bayesian nonparametric models with applications in science, engineering, finance and medicine.

Slides