The application of categorical embedding and spatial-constraint clustering methods in nested GLM model

Generalized linear model (GLM) is a popular modeling choice for pricing non-life insurance policies. However, high-cardinality categorical insurance data presents significant challenges for these GLM ratemaking models. Additionally, insurance regulators often require rating territories, which are clusters of insurance policies' geographic locations used for setting insurance rates, to meet certain standards such as credibility, contiguity, and cardinality. To address these challenges, this thesis proposes a nested GLM framework that integrates GLM with a neural network model with categorical embedding layers and spatial-constraint clustering models. The nested GLM satisfies regulatory requirements, and enhances the model's predictive power, while maintaining the interpretability from the (generalized) linear form. The construction of a nested Poisson GLM is presented in this thesis. Its performance is demonstrated using a real-life Brazil auto insurance data.