Abstract

This project's primary goal is to forecast medical insurance costs using SVM, XGBoost, Random Forest, and Linear Regression machine learning algorithms. The process involves creating a dataset from scratch that is similar to real health data, preparing it, applying various algorithms to it, and evaluating its performance. The Target variable in the dataset is medical insurance expenses, while other parameters include age, BMI, children, gender, smoking habits, region, blood pressure, cholesterol levels, and diabetes. Min-Max scaling is used to normalize numerical characteristics, whereas one-hot encoding is used for categorical variables. To optimize performance, the models are trained using the pre-processed data. Model evaluation uses evaluation metrics like R-squared and Mean Squared Error (MSE). Cross-validation assures consistency in model prediction. The results contain data extracted from best-fit lines illustrating the predictions generated by every program. The program provides an in-depth review of medical insurance cost forecasts for the benefit of healthcare and insurance decision-makers.