Bayesian and Generative Probabilistic Modeling in Marketing

The course aims to provide an in-depth understanding of modern Bayesian and probabilistic machine learning approaches for modeling consumer data to inform firm decision making. The course also aims to provide students with the necessary computational skills in the use of modern computation methods and frameworks. These include modern variants of Markov chain Monte Carlo methods based on Hamiltonian Monte Carlo and Sequential Monte Carlo techniques as well as recent variational Bayesian approaches for scalability and speedy computation. The course will also cover the analysis of panel data and multilevel data structures using hierarchical Bayesian methods and econometric models used in empirical modeling of marketing/business phenomena. The course is primarily targeted for PhD students. Students are required to have a good quantitative background with an understanding of the basics of probability, statistics, and econometrics. In addition, students should be familiar with a programming environment such as Python or Julia and willing to learn JAX, TensorFlow, or PyTorch.