Seminar & Thesis
Seminar
The seminar allows students who took both Data Science I & II to dig deeper on topic that they have previously encountered in class or elsewhere. Such topics from the past include:
- Tree models: conditional inference trees, panel data,
- Strategies for unbalanced data
- Dimensionality reduction: UMAP
- Clustering: HDBSCAN
- Neural Network Architectures: autoencoders, RNN
- Causal Inference: Double Machine Learning
Students prepare a 10-15 page paper and/or a notebook that illustrates the topic. The the level of technical details and the focus on either of these two resources depends on the specific topic.
The topic will be presented at the end of the semester. Key criterion for a successful presentation is the ability to convey the topic to the other participants in a simple fashion, with simple examples and the ability to answer questions. You have 30 minutes for your presentation followed by 30 minutes of discussion, questions and commentary from the supervisor.
Thesis
I am happy to supervise or co-supervise theses that involve data analyses or employ methods from my modules. If you choose me as your main supervisor, you should either write about a technical topic such as a method - possibly garnished with a small application - or about a topic from the field of health economics.
Please note that a thesis is no side hustle. You should complete an internship or half a dozen classes with exams in parallel. Your thesis requires focus for 8 weeks straight.
Please use this template for your seminar paper and/or thesis.
Template.docx
(75,7 KB) vom 26.03.2026