MACHINE LEARNING TECHNIQUES FOR ACTUARIAL RISK MANAGEMENT
International Journal of Sciences and Innovation Engineering · 02.11.2026 · 3 (11) · 119-128
DOI: 10.70849/ijsci03112615959 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Sciences and Innovation Engineering, online erschienen am 02.11.2026. Verfasst von James A. Otunomeruke und 1 weiteren Autor:innen.
Abstract (Originalsprache)
Proper risk assessment of the insurance companies in terms of financial stability is only possible through accurate actuarial assessment of risk, but the traditional statistical models are prone to the failure of dealing with high-dimensional, non-linear data. The current work is an evaluation of three machine learning algorithms which include Gradient Boosting (GBM), Random Forest (RF) and Support Vector Machines (SVM) to predict the severity of insurance claims using the Allstate Auto Insurance dataset (around 300,000 cases, 188 features). Data was pre-processed in detail, with missing values being imputed, categorical variables being encoded, and values being scaled and feature-engineered, and hyperparameter optimization then performed and followed by cross-validation to provide a strong model evaluation. Results indicate that GBM consistently outperforms RF and SVM, achieving the lowest prediction errors (MAE = 1,128; RMSE = 1,894) and the highest explanatory power (R2 = 0.882). Random Forest presents comparable competitive accuracy and a slightly larger error whereas SVM is not as strong with high-dimensional and complex data. Ensemble models, in particular, GBM as a model, is used well to capture non-linear interactions and extreme values of claims hence offer more correct and reliable predictions compared to the conventional models. The results prove that machine learning, particularly Gradient Boosting, when applied to actuarial models, can improve the efficiency of claims prediction, risk segmentation, and underwriting. These methods can allow being proactive and data-driven in the management of risks, which will lead to better operational and financial balance in the insurance industry.
Zitieren
APA
Otunomeruke, J. A., & Osolo, M. O. (2026). MACHINE LEARNING TECHNIQUES FOR ACTUARIAL RISK MANAGEMENT. International Journal of Sciences and Innovation Engineering, 3(11), 119-128. https://doi.org/10.70849/ijsci03112615959
Vancouver
Otunomeruke JA, Osolo MO. MACHINE LEARNING TECHNIQUES FOR ACTUARIAL RISK MANAGEMENT. International Journal of Sciences and Innovation Engineering. 2026;3(11):119-128. doi:10.70849/ijsci03112615959
BibTeX
@article{Otunomeruke2026,
author = {Otunomeruke, James A. and Osolo, Moses O.},
title = {MACHINE LEARNING TECHNIQUES FOR ACTUARIAL RISK MANAGEMENT},
journal = {International Journal of Sciences and Innovation Engineering},
year = {2026},
volume = {3},
number = {11},
pages = {119--128},
doi = {10.70849/ijsci03112615959}
}
EUVIMED