2-3 hours per week
Bayesian Statistics is a captivating field and is used most prominently in data sciences. In this course we will learn about the foundation of Bayesian concepts, how it differs from Classical Statistics including among other Parametrizations, Priors, Likelihood, Monte Carlo methods and computing Bayesian models with the exploration of Multilevel modelling.
This course is divided into two parts i.e. Theoretical and Empirical part of Bayesian Analytics. First three weeks cover the Theoretical part which includes how to form a prior, how to calculate a posterior and several other aspects. Rest of the weeks will cover the empirical part which explains how to compute Bayesian modelling. Completion of this course will provide you with an understanding of the Bayesian approach, the primary difference between Bayesian and Frequentist approaches and experience in data analyses.
Target Audience: Business school students, working professionals and anyone who is interested in learning Bayesian statistics with the basic knowledge of mathematics.
Pulak Ghosh is Professor in the Decision Sciences Area at IIMB. His key specializations are in intersection of Big data, Machine learning, Artificial Intelligence and its use in Economics, Finance, Policy and Social Value Creation. He did serve in the editorial board of Journal of the American statistical Association, Journal of the Royal statistical Society and currently serves in the editorial board of Biometrics.
Based on his outstanding and innovative contribution to research, the International Indian Statistical Association awarded him with the “Young Scientist Award” in 2011. Govt of India awarded him the prestigious CR Rao award in 2015 and Econometric Society awarded him the Mahalanobis Award in 2016.
Prior to joining IIMB, he served as Associate Director, Novartis Pharmaceuticals, USA, Assistant Professor, Georgia State University, and Associate Professor at Emory University, USA. He is a visiting faculty at several institutes of international repute.
Week 1: What is Bayesian Statistics and How it is different than Classical Statistics
- Foundations of Bayesian Inference
- Bayes theorem
- Advantages of Bayesian models
- Why Bayesian approach is so important in Analytics
- Major densities and their applications
Week 2: Bayesian analysis of Simple Models
- Likelihood theory and Estimation
- Parametrizations and priors
- Learning from binary models
- Learning from Normal Distribution
Week 3: Monte Carlo Methods
- Basics of Monte carol integration
- Basics of Markov chain Monte Carlo
- Gibs Sampling
Week 4: Computational Bayes
- Examples of Bayesian Analytics
- Introduction to R and OPENBUGS for Bayesian analysis
Week 5: Bayesian Linear Models
- Context for Bayesian Regression Models
- Normal Linear regression
- Logistic regression
Week 6: Bayesian Hierarchical Models
- Introduction to Multilevel models
- Computation in Hierarchical Models
- Understand the necessary Bayesian concepts from practical point of view for better decision making.
- Learn Bayesian approach to estimate likely event outcomes, or probabilities using datasets.
- Gain “hands on” experience in creating and estimating Bayesian models using R and OPENBUGS.