«Actuarial Data Science Après-Midi»

Topic                      Tree-like pairwise interaction network (PIN)

Date & Time          Wednesday 15.10.2025, 18:00 to 19:00
Location                 Hotel St. Gotthard
SAA CPS                1
# Participants    35

Abstract
Modeling feature interactions in tabular data remains a key challenge in predictive modeling, for example, as used for insurance pricing. We present the Tree-like Pairwise Interaction Network (PIN), a novel neural network architecture that explicitly captures pairwise feature interactions through a shared feed-forward neural network architecture that mimics the structure of decision trees. PIN enables intrinsic interpretability by design, allowing for direct inspection of interaction effects. Moreover, it allows for efficient SHapley's Additive exPlanation (SHAP) computations because it only involves pairwise interactions. We highlight connections between PIN and other established models. Empirical results on the popular French motor insurance dataset show that PIN outperforms both traditional and modern neural networks benchmarks in predictive accuracy, while also providing insight into how features interact with each another and how they contribute to the predictions.

Mario Wüthrich
Mario is a Professor in the Department of Mathematics and the Director of Actuarial Studies at ETH Zurich. He did his PhD in Mathematics at ETH Zurich in 1999. An Actuary SAA since 2004, he served on the board of the Swiss Association of Actuaries from 2006 to 2018. Since 2018, he has been the Editor-in-Chief of the ASTIN Bulletin, and since 2025, he has served as Senior Scientific Advisor at InsureAI.