ENHANCED MEDICAL INSURANCE COST PREDICTION WITH MACHINE LEARNING

Abstract

The paper proposes an intelligent prediction model for estimating the annual medical insurance costs of individuals based on demographic, health, and policy information. It proposes the utilization of data processing, feature scaling, and group learning in tandem with a Random Forest Regressor. The data includes attributes such as age, BMI, income, smoking habits, and policy details. A standardized numerical data processing pipeline and One Hot Encoder for categorical encoding form the backbone. The R² for the trained model is 89.58%, demonstrating that it explains a large part of the variance in insurance charges. Further, a web-based user interface using Streamlit has been created, through which a user can input values and instantaneously get their predicted annual medical costs in Indian Rupees.