Humboldt-Universität zu Berlin - School of Business and Economics

General Information

Course description

This seminar examines how statistical and machine learning methods tackle the challenge of missing and partially observed data. We will start by introducing key concepts of missingness, using examples from real-world applications such as survey nonresponse and unlabelled data. Building on this, we will explore how causal inference and semi-supervised learning can be understood as special cases of learning under missing data. In the following sessions, students will present and discuss recent research papers on imputation, causal inference with missing data, and semi-supervised learning, followed by group discussions. Toward the end of the seminar, each student will submit a short, paper-style report based on the research paper they presented. This report will be formally assessed.

Recommended prior lectures or prior knowledge: Participants should be familiar with core concepts in probability and statistical inference (e.g. Statistical Inference or equivalent). Prior coursework in computational statistics, econometrics, or machine learning is helpful but not required.

Registration

Note that the number of participants is usually limited and a registration is necessary. Please refer to the most recent course catalogue on Agnes for more information.