Bayesian Modeling in Data Science: a Practical Approach using R and BUGS/JAGS

Pablo-Emilio Verde

In recent decades, Bayesian modeling has gained popularity in applied statistics and data science. One of the primary reasons for this success has been the development of free and open-source statistical software for performing MCMC (Markov Chain Monte Carlo) computations. These computational techniques integrate Artificial Intelligence (AI) with computer simulations, enabling Bayesian modeling to accurately represent the complexity of the data (e.g., hierarchical structures, missing data, outliers).

This course aims to give a practical introduction to Bayesian data analysis using the statistical software R and the computer simulation software BUGS/JAGS.

The following topics will be covered during the course:

Day 1 (19.5.):

• Bayesian uncertainty quantification
• Bayesian analysis of single-parameter models
• MCMC with R and BUGS/JAGS

Location:

• Seminar Room 10:00 - 12:00
• room M3 12:00 - 13:00
• Seminar Room 14:00 - 18:00

Day 2 (20.5.):

• Posterior predictions: Bayesian model checking of multiple-parameter models
• Bayesian approaches for missing data
• Bayesian regression models: Ridge Regression, LASSO and variable selection, Robustification of nonlinear models, Risk analysis and logistic regression

Location:

• Seminar Room 9:00 - 13:00
• room MP2 14:00 - 15:00
• room M3 15:00 - 16:00

Day 3 (21.5.):

• Machine learning and Bayesian Adaptive Regression Trees (BART)
• Bayesian Nonparametrics
• Bayesian hierarchical modeling

Location:

• room M4 9:00 - 10:00
• Seminar Room 14:00 - 18:00