Winter 2021/22: Bayesian Statistics I

Contents

Bayesian thinking differs from frequentist statistics in its interpretation of probability and uncertainty. It complements the existing statistical toolbox with powerful methods for simulation and inference. The lectures Bayesian Statistics I and II aim to familiarize the students to the Bayesian approach. The first part deals with the theoretical fundamentals and the principles of estimating, testing, forecasting and model assessment. In addition, Bayesian regression concepts and computer-​intensive simulation methods such as Markov chain Monte Carlo (MCMC) are introduced. The second part complements and deepens these topics, for example by Bayesian nonparametric density estimation, Bayesian model choice and Approximate Bayesian Computing (ABC).

General information

Lecturers: Prof. Dr. Christiane Fuchs (lectures), Houda Yaqine (exercises)

Type: Lecture with (optional but recommended) exercises

Study achievements (Studienleistungen): Study achievements for the exercise classes can be fulfilled by preparation and one submission of exercise sheets.

Recommended prerequisites: Good knowledge of statistics (esp. (conditional) densities/probabilities, likelihood inference, regression) and R

Module allocation: see eKVV (lecture) and eKVV (exercises)

Dates: The lectures and exercise classes take place in person at the university on Thursdays and Fridays as listed in the following. Please check this page regularly for updates!

Literature

• Lee: Bayesian Statistics. Wiley, 4th edition.
• Gelman et al.: Bayesian Data Analysis. CRC Press, 3rd edition.
• Held & Sabanés Bové: Applied Statistical Inference. Springer.

Material

Lecture slides, exercise sheets and further material will be made available via LernraumPlus.

This class is supported by DataCamp, a learning platform for data science. Members of this class can access all courses for free. The invitation link is available through LernraumPlus.