Marjolein Fokkema joins RC Trust

RC Trust welcomes Marjolein Fokkema as a new professor. Since August 2026 she holds the Chair of Computational Statistics with Applications in Psychology at the Department of Statistics at TU Dortmund University and has been affiliated with RC Trust.

Her research lies at the intersection of statistical modeling, machine learning, artificial intelligence, psychological assessment, and psychometrics. At its core is a fundamental question: How can researchers learn from data how human behavior works?

Machine learning and AI systems are often promoted for their predictive performance. Yet strong predictions alone do not necessarily lead to scientific understanding. Many models are highly complex, difficult to interpret, and unable to provide reliable information about the uncertainty surrounding their results.

Fokkema and her group develop methods that address this gap. A particular focus is on decision trees and ensemble methods. Decision trees are based on a simple principle: Data points are repeatedly divided into subgroups whose members become increasingly similar. The resulting models can often be represented as a sequence of understandable decisions.

A single tree, however, may be too simple to capture the nuances of human behavior. Combining many trees can make predictions more flexible and accurate, but it also makes the model more complex. This creates a central challenge: How can researchers retain the predictive strengths of these methods while still drawing transparent and reliable scientific conclusions?

“Making tree and ensemble methods understandable and quantifying their uncertainty” is therefore a key focus of Fokkema’s group.

For her, trustworthy data science also means being open about the limits of prediction. Machine learning methods are not always clearly superior to simpler approaches such as linear regression. Even the best models leave much of human behavior difficult to predict.

“The psychologist in me is actually quite happy about that,” Fokkema says.

Before joining TU Dortmund University, Fokkema worked at Leiden University for more than eleven years, first as Assistant Professor and later as Associate Professor. She previously completed her doctoral research and taught at VU University Amsterdam and spent six months as a visiting researcher at the University of Zurich.

The interdisciplinary environment was a major reason for her move. “Psychologists, computer scientists, and statisticians working together is the perfect environment for me,” she says.

At RC Trust, Fokkema plans to develop methods for scientific inference with machine learning and AI, collaborate across disciplines, and strengthen connections with Leiden University. She also develops open-source statistical software and explores creative science communication through her band project, Mary-Jo and the Support Vector Machines.