ľ¹Ï¸£ÀûÓ°ÊÓ

This course provides students with a strong foundation of statistics in data analysis: from refreshing common modelling techniques and their assumptions to exploring fundamental principles of statistical learning, such as prediction, causation, and the bias-variance trade-off. The aim is to provide students with a broad and critical perspective on both the potential and the limitations of common statistical approaches.

To be able to understand and apply modern statistical methods, students need a strong mathematical basis. Starting from basic calculus, matrix algebra, and probability, this course introduces several fundamental concepts in mathematical statistics, such as unbiasedness, the likelihood, estimation, probability densities, and hypothesis testing. By the end of the course, students are equipped with the tools and skills to approach statistical problems with mathematical rigour.

This course introduces the basics of modern programming in R, using the tidyverse suite of packages. Students will also focus on statistical inference and computational methods central to data analysis: topics such as simulation, resampling, and algorithmic techniques provide simulating content for developing advanced programming skills. Real-world datasets and applied exercises form the core of the learning experience.

This course consists of two parts. The first part focuses on the challenge of identifying causal effects when randomized controlled trials are not feasible. Building on fundamental theoretical frameworks such as potential outcomes and directed acyclic graphs (DAGs), students explore modern methods of causal inference. The course provides practical experience in hands-on lab sessions using R bridging theory and application for causal research questions. 

The second part deals with prediction and forecasting, which are core tasks in science and beyond. Students will learn to justify, implement, and evaluate various modern machine learning techniques for supervised learning problems. Topics include random forest, tree-based boosting methods, and (deep) neural networks. The course is framed in the statistical learning context, with an eye for data-generating processes, sampling, and the trade-off between bias and variance.

This course consists of two parts. The first part focuses on population inference as a key goal of almost any scientific project: How can we use data to produce knowledge about the world around us? Students are guided through the complete statistical process, from sampling to modern estimation methods for drawing inferences from complex surveys and multi-source data. Topics include sampling, handling nonresponse, weighting, and advanced techniques such as inference for non-probability samples, data integration, and small-area estimation. 

The second part of this course focuses on hands-on group projects, giving students the skills to handle, wrangle, and model the large diversity of data formats that exist in science and beyond. From databases to geospatial information systems, from sensor data to high-frequency longitudinal sampling; the data science world has moved on from regular spreadsheets, and modern statisticians need to be able to handle diverse data types.

This course is divided into two complementary parts. The first introduces multilevel models, which are regression models designed for data with a nested structure. Beyond the basic multilevel model, this course will also move to more advanced topics such as analysing longitudinal data. 

The second part focuses on structural equation modelling (SEM), which is a popular framework that integrates measurement and path models. This framework encompasses factor and path analyses, complex mediation analyses, and longitudinal analyses. In both parts, practical lab sessions connect theory to practice.

In the first part of this course, you will learn the basics of Bayesian thinking and get hands-on experience by building and programming your own Bayesian samplers in R. You will explore how Bayesian methods work in practice, from setting priors to estimating posteriors, and apply them to real research questions. The course also introduces Bayesian model evaluation and model checking, essential tools for assessing and comparing models in applied research. 

In the second part, you will study tools designed for studying and modelling attributes that cannot be directly measured (i.e., latent variables). These techniques have their roots in psychology, where many attributes are latent (e.g., intelligence), hence 'psychometrics'. This course covers several approaches, including classical test theory, factor analysis, and item response theory.

This course deals with the relationship between medical research questions and the study designs used to answer them. Students will examine observational and experimental designs such as cohort studies, case-control studies, and clinical trials, as well as statistical methods such as survival analysis and meta-analysis. The emphasis is on learning how to translate a research question into a study design, statistical plan and analysis, and how to communicate results effectively in biomedical contexts.