A Review of Emerging Research Trends based on Machine Learning Models for Insurance Premium Prediction

Abstract

Insurance premium prediction is a fundamental task in the insurance industry, as it enables insurers to accurately assess risk, determine fair pricing, and maintain financial sustainability. The relations that exist, perhaps in the majority of insurance industry data, are mostly nonlinear, and hard to accurately model with traditional actuarial methods. Therefore, machine learning (ML) techniques have been developed as efficient alternatives in order to provide improvement of the accuracy of premium prediction by using data-driven risk assessment. This survey provides an up-to-date overview of the latest developments in the field of insurance premium prediction based on ML algorithms, including the popular methods Support Vector Regression (SVR), Random Forest (RF), Linear Regression (LR), Gradient Boosting Machine (GBM), eXtreme Gradient Boosting (XGBoost), and Long Short-Term Memory (LSTM). Moreover, the paper discusses the research trends, including Explainable Artificial Intelligence (XAI), Large Language Models (LLM), telematics-based pricing, AI-guided personalized insurance, and MLOps, which leverage these advancements to improve transparency, automation, and decision support in premium prediction. Additionally, the survey highlights some of the challenges faced, including data quality, model interpretability, security and regulatory compliance, and summarizes the literature of the past few years and proposes future research directions. In this review, researchers and practitioners get a comprehensive overview of existing methodologies, technological advances, and open challenges in machine learning for insurance premium prediction.