Detection of Interacting Variables for Generalized Linear Models via Neural Networks
This work addresses the time-consuming and expert-dependent process of finding interactions in GLMs, particularly for insurance companies, offering an incremental improvement in automation and speed.
The paper tackles the problem of automating the detection of interacting variables in generalized linear models (GLMs) to improve predictive power, using neural networks and a model-specific method that is computationally faster than traditional approaches like Friedman H-Statistic or SHAP values, with results demonstrated on artificially generated and open-source data.
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 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.