University of Twente Student Theses

Login

Beyond Generalized Linear Models : Advancing Insurance Pricing through Interpretable and Explainable Machine Learning

Reil, J.P.C. (2024) Beyond Generalized Linear Models : Advancing Insurance Pricing through Interpretable and Explainable Machine Learning.

[img] PDF
11MB
Abstract:This thesis explores the integration of machine learning (ML) techniques into non-life (car) insurance pricing, traditionally handled by generalized linear models (GLMs). While GLMs offer transparency and meet regulatory requirements, the potential for enhanced predictive accuracy has made the insurance industry to explore ML alternatives. This research examines whether ML can improve premium pricing predictions while maintaining the interpretability and explainability essential for industry adoption. The study applies various ML algorithms, comparing them against traditional GLMs using various performance metrics for both severity and frequency data. To address concerns about the transparency of ML models, XAI techniques such as SHAP and LIME are employed. Results show that ML models outperform GLMs in predictive power for both severity and frequency claims data. However, ensuring model explainability, especially in regulatory and compliance contexts, remains a challenge, as highlighted through interviews with industry stakeholders. The research proposes a hybrid approach, where ML models complement GLMs, combining their strengths to improve accuracy without sacrificing explainability and interpretability. This study contributes to the existing body of research on AI-driven insurance pricing.
Item Type:Essay (Master)
Clients:
Triple A - Risk Finance, Amsterdam, Netherlands
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:83 economics
Programme:Business Information Technology MSc (60025)
Link to this item:https://purl.utwente.nl/essays/103673
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page