The Implementation of Machine Learning Algorithms in Non-Life Insurance : A Literature Review
Abstract
The non-life insurance sector is extremely competitive, and insurance companies need to create pricing that is adjusted to market requirements. Typically, insurers use generalized linear models to model pure premium. These traditional models have limitations that impose constraints on the structure of the modeled risk and on the interactions between risk explanatory variables, which can lead to a biased estimate of the insurance premium. To overcome these limitations, actuaries opt for efficient algorithms, known as machine learning models, which have demonstrated their ability to extract dependency structures and peculiarities between data.
The use of these methods in non-life insurance has grown significantly in recent years, due to their ability to estimate the pure premium in a way that is accurate and tailored to each policyholder's risk profile. Similarly, machine learning algorithms offer advantages in terms of risk management, fraud detection and policy underwriting automation.
In this article, we examine the development of machine learning algorithms in non-life insurance, identifying the limitations of generalized linear models as well as the various tools used to evaluate the performance of the models adopted.
Keywords: underwriting, GLM, pure premium, machine learning, statistical learning.
Classification JEL: C60, G22
Paper type: Theoretical Research
Downloads
Copyright (c) 2024 Safaa MOKDAR, Tarek ZARI
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.