Comparative Analysis of Machine Learning Models for Health Insurance Premium Prediction Using R

Abstract

This study explores methodologies for forecasting health insurance premiums, focusing on predictive accuracy and reliability. Using a dataset with variables such as age, gender, BMI, and diseases, we apply multiple techniques-including the K-Nearest Neighbors (KNN) algorithm, voting methods, and other machine learning algorithms-to predict premiums. A comparative analysis highlights each method's strengths and limitations, offering insights into which approach provides the most accurate and practical predictions. The findings aim to guide insurers in selecting effective forecasting methods to enhance premium pricing strategies and improve risk management.