On Friday 3 October 2025, PhD candidate Daniël Touw (Erasmus School of Economics) will defend his dissertation, in which he presents new algorithms for uncovering structure in large and complex datasets. His research develops scalable, interpretable methods for clustering, network estimation, and predictive modelling, essential tools for modern data science.
At the heart of the dissertation is a new algorithm for convex clustering, which groups similar items orders of magnitude faster than existing methods and scales to datasets with over one million observations. This approach is further extended to convex biclustering, enabling the simultaneous clustering of both observations and variables with computational complexity that grows only linearly with dataset size.
Touw’s work also introduces the clusterpath estimator of the Gaussian graphical model (CGGM), a method that efficiently detects block-structured networks of variables using a fast coordinate descent algorithm. Applications in both simulated and empirical settings demonstrate strong performance in identifying clusters of interdependent variables.
A final contribution is the classifier chain network for multi-label classification. By jointly modelling dependencies between multiple outcomes, this method achieves improved predictive accuracy in diverse applications. Together, these advances mark important progress in making powerful data analysis techniques feasible at massive scale, with significant benefits for both scientific research and applied machine learning.
Public defence
The public defence will take place on Friday 3 October 2025 at 1pm. Please note that the doors will close promptly at the start of the ceremony. A livestream link will be made available a few days before the event.
- More information
For more information, please contact Ronald de Groot, Media & Public Relations Officer at Erasmus School of Economics: rdegroot@ese.eur.nl, mobile +31 653 641 846.
