Detection of interacting variables for generalized linear models via neural networks

Publikation: Bidrag til tidsskriftTidsskriftartikelForskningfagfællebedømt

The quality of generalized linear models (GLMs), frequently used by insurance companies, depends on the choice of interacting variables. The search for interactions is time-consuming, especially for data sets with a large number of variables, depends much on expert judgement of actuaries, and often relies on visual performance indicators. Therefore, we present an approach to automating the process of finding interactions that should be added to GLMs to improve their predictive power. Our approach relies on neural networks and a model-specific interaction detection method, which is computationally faster than the traditionally used methods like Friedman’s H-Statistic or SHAP values. In numerical studies, we provide the results of our approach on artificially generated data as well as open-source data.

TidsskriftEuropean Actuarial Journal
StatusE-pub ahead of print - 2024

Bibliografisk note

Funding Information:
We acknowledge the support of ERGO Center of Excellence in Insurance, funded by the ERGO Group AG. We thank Kay Adam for providing the data as well as for valuable suggestions and Frank Ellgring for the opportunity to gain practical insights in actuarial pricing at Global P &C Pricing Department at ERGO Group AG. We acknowledge the support of Noel Stein, Samarth Mehrotra, Mario Ponce-Martinez, and Yichen Lou in the preparation phase of this project.

Publisher Copyright:
© 2023, The Author(s).

ID: 372718973