Publication

Abstract : This research aims to estimate the health insurance claim costs using a machine learning algorithm with given potential health risks factors including age, sex, BMI, smoking status and city. To benchmark the prediction accuracy for various algorithms selected in this analysis, a health insurance dataset obtained from Kaggle is used, details of which are given as follows: It involves activities such as data pre-processing, feature engineering, model selection along with using measures like R-squared (R2ˆ) and Mean Absolute Error. The predictive accuracy results indicated that Random Forest rendered the best results among all the tested models with an accuracy of 96.7% in modeling the health data due to its rich ability to generate various relationships. This research also focuses on the capability of machine learning in cost forecasting for insurance companies that would improve on the pricing strategies among other aspects of financial management.