During the VVSOR Annual Meeting on the 21th March 2024, the Van Zwet Award for best PhD-thesis has been handed out. This year, in 2024, there was one winner, Richard Post, and one honorable mention, Wouter Kool. Below we share the jury report of the chair of the jury, Jelle Goeman.

Jury Report 2024

This year we received a fair number of high quality nominations for the van Zwet award. We note, as in previous years that the nominations were skewed towards from Mathematical Statistics and Operations Research. As every year I repeat my request to promotors in applied statistics: be less modest and nominate your excellent theses!

All submitted theses were very good but also very different, and as a jury we had a long discussion trying to rank the unrankable. In the end we agreed on a single winner, but we would like, first, to give an honorary mention to a strong contender.

The thesis “Learning and Optimization in Combinatorial Spaces with a Focus on Deep Learning for Vehicle Routing” by Wouter Kool, and supervised by Prof. dr. M. Welling, dr. H.C. van Hoof and prof. dr. J.A.S. Gromicho. This is a exiting work at the intersection of statistics, machine learning and operations research, and the jury was excited to see these fields brought together in a very creative way. In the first part of the thesis, Wouter Kool develops new methods for vehicle routing using reinforcement learning. In the second part, he develops novel statistical techniques for sampling without replacement from the search space, in order to make the reinforcement learning more efficient. The work has already had a large impact, with the chapter “Attention! Learn to solve routing problems”, published 2019, already cited over 1200 times. Let us congratulate Wouter with his thesis.

The winner of this year’s Willem R. van Zwet award is Richard Post, with his thesis “Causal Effect Heterogeneity: statistical formalization and analysis of the individual causal effect”, supervised by Edwin van den Heuvel and Hein Putter. Richard has chosen a rather daring subject for his thesis. The subject of individual causality, or causal effect heterogeneity. The subject is controversial, since individual causal effects seem to require individual counterfactuals, which by definition are never observed. However, Richard describes models and assumptions under which it is possible to see that causal effects are not the same for different individuals, and develops methods to chart the resulting heterogeneity. This causal heterogeneity is relevant in cross-sectional but also in longitudinal outcomes and survival analysis, which he explores.

The jury was impressed by the broad scope encompassing both theory and applications. The exposition is very clear. The jury found Richard’s work to be of the highest scientific quality. It has mathematical rigor as well as intuitive explanations, strong modelling, it is conceptually highly innovative, and illustrates all this with relevant applications. Moreover, and we find this very important, the thesis is also very well written.

We congratulate Richard on a well-deserved award!

Learn more about the different VVSOR awards and the jury here