Global Journal of Economic and Finance Research

This study explores the integration of actuarial science and artificial intelligence through a hybrid framework combiningoptimization techniques and deep learning to enhance actuarial risk modeling. Using a Moroccancase study, we leverage linear programming for financial decision-making and arti- ficial neural networks (ANN) for predicting and analyzing actuarial risks under uncertainty.

The methodology is applied to real-world data from the Moroccan insurance sector, demonstrating its practical relevance for improving long-term financial planning and risk evaluation.

Experimental results show that the ANN model achieves a normalized Mean Absolute Error (MAE) of 0.625, outperforming traditional Generalized LinearModels (GLMs) by approximately 15% in pre- dictive accuracy. Risk assessment metrics further revealimprovements, with the 95% Value at Risk (VaR) and Tail Value at Risk (TVaR)reduced by 9.4% and 13.9%,respectively, compared to Poisson andNegative Binomial GLMs. These gains highlight the ANN’s capacity to capture complex nonlinear relationships and heteroscedasticity in insurance claim data.

Moreover, linear programming integration enables optimized premium pricing and reserve allocation under multiple regulatory and economic constraints.

This hybrid approachfosters a more robust and adaptive actuarial process, equipping insurersto bet- ter manage uncertainty and extreme risk scenarios.

Overall, this work modernizes actuarial practices by introducing intelligent, data-driven modelsthat enhance predictive performance and strategic decision-making, contributing to sustainable financial stability in emerging markets such as Morocco.

automobile insurance, premium pricing, generalized linear models (GLM), credibility theory, risk distribution.

1. Anisetti, M., Ardagna, C. A., Bena, N., & Foppiani, A. (2021). An assurance-based risk management framework for distributed systems. 2021 IEEE International Conference on Web Services (ICWS), 482–492.
2. Antonio, K., & Valdez,E. A. (2012). Statistical conceptsof a priori and a posteriori risk classification in insurance. ASTIN Bulletin.
3. Bühlmann, H. (1967). Experience rating and credibility. ASTIN Bulletin, 3(3), 199–207.
4. Bühlmann, H., & Gisler,A. (2005). A coursein credibility theoryand its applications. Springer.
5. Cai, J., & Tan, K. S. (2000). Conditional tail expectation for elliptical distributions. Insurance: Math- ematics and Economics, 27 (3), 345–356.
6. Chapman, P., Clinton, J., Kerber, R., Khabaza, T., Reinartz, T., Shearer, C., & Wirth,R. (2000). Crisp-dm 1.0: Step-by-step data mining guide. SPSS Inc.
7. Charpentier, A. (2020). Data sciencefor actuaries. Springer.
8. Chen, T., & Guestrin, C. (2016). Xgboost: A scalable tree boosting system. Proceedingsof the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.
9. de Jong, P., & Heller, G. Z. (2008).Generalized linear models for insurance data. Cambridge University Press.
10. Denuit, M., et al. (2007).Actuarial modelling of claimcounts: Risk classification, credibility and bonus-malus systems. Wiley.
11. El Attar, M., et al. (2019). Application of neural networks to automobile insurance pricing. Revue Marocaine de Rechercheen Management et Marketing.
12. Frees, E. W. (2010).Regression modeling with actuarial and financial applications. Cambridge Uni- versity Press.
13. Gerber, H. U. (1972). Games of economic survival with insurance applications. ASTIN Bulletin, 6(2), 111–135.
14. Gong, Y., Li, Z., Milazzo, M., Moore, K., & Provencher, M. (2018). Credibility methods for individual life insurance. Risks, 6(4), 144.
15. Hastie, T., Tibshirani, R., & Friedman, J. (2009). The elements of statistical learning. Springer.
16. Herzog, T. N. (1999).Introduction to credibility theory. Actex Publications.
17. Janssens, D., Wets, G., Brijs, T., Vanhoof, K., & Vanthienen, J. (2006). Integrating neural networks for car insurance fraud detection. Expert Systems with Applications, 29(3), 653–666.
18. Kaas, R., Goovaerts, M., Dhaene, J., & Denuit, M. (2008). Modern actuarial risk theory: Using r [Application directe de VaR et TVaR dans un contexte actuariel avec implémentation R]. Springer.
19. Klugman, S. A. (1992). Loss models and their estimation. Insurance: Mathematics and Economics, 11(2), 105–121.
20. Klugman, S. A., Panjer, H. H., & Willmot, G. E. (2012).Loss models: From data to decisions (4th).Wiley.
21. Liu, Y., & Shih, Y.-H. (2011). Integrating data mining and behavioral scoring for risk assessment of insurance customers. Expert Systems with Applications, 38(6), 6551–6560.
22. Mahler, H. C., & Dean, C. G. (1999).Credibility (chapter 8) [Excellent chapitreappliqué pour les praticiens]. Foundations of Casualty ActuarialScience, 4th Edition.
23. McCullagh, P., & Nelder, J. A. (1989). Generalized linear models (2nd). Chapman; Hall/CRC. 
24. Ohlsson, E., & Johansson, B. (2010). Non-life insurancepricing with generalized linear models (Vol. 174). Springer.
25. Rouyan, Y., & Amrani,S. (2020). Étude des déterminants de changement d’assureur automobile au maroc. Revue Marocaine de Management, 10(1), 77–92.
26. Shearer, C. (2000). The crisp-dm model: The new blueprint for data mining. Journal of Data Ware- housing, 5(4), 13–22.
27. Tan, K. S., & Weng, C. (2006). Value at risk and conditional tail expectation for actuarial risks: A comparative study. North AmericanActuarial Journal, 10(4), 100–114.
28. Taylor, G. C., & Ashe, D. (1983). Credibility, exposure and risk measures in insurance. Journal of the Royal Statistical Society: Series C (Applied Statistics).
29. Wang, S., Young, V. R., & Panjer, H. H. (1998). Axiomatic characterization of insurance prices [Lien entre stop-loss, TVaR et principes actuariels de tarification]. Insurance: Mathematics and Economics, 21(2), 173–183.
30. Wüthrich, M. V. (2018).Machine learning in insurance: Theory and case studies. Springer. 
31. Zehnwirth, B. (2002). The negative binomialdistribution in insurance. Insurance: Mathematics and Economics, 30(1), 33–47.