Health Insurance Recommendation System using Optimized Grid Search and Regression Models

Introduction: Health insurance schemes help cover medical expenses by distributing financial risk among many individuals. With various insurance options available, choosing the right provider and predicting costs can be challenging. Predictive modeling and machine learning techniques play a important role in analyzing past data, identifying patterns in customer behavior, and supporting informed decision-making for new insurance plans. Objectives: The main aim of this research is to assist individuals in selecting appropriate medical insurance providers and estimating associated costs using predictive models. By leveraging historical data, the study seeks to improve cost prediction accuracy and enhance decision-making in the health insurance sector. Methods: This study utilizes medical provider datasets along with cost prediction data to develop predictive models. A total of 12 regression classifiers are applied to analyze the data. To optimize performance, Grid Search Cross-Validation is used for fine-tuning the models. This ensures better accuracy and reliability in predicting insurance costs. Results: From analysis, X-Gradient Boost, Random Forest, and Extra Trees models demonstrated the highest accuracy. These models achieved R² scores greater than 98%, indicating their effectiveness in capturing the relationship between input features and insurance costs. The Extra Trees model perform well with an R² score of 0.99 during training and 0.88 during testing. Additionally, these models provide low Root Mean Squared Error (RMSE) values, confirming their reliability in making precise predictions. Conclusions: The findings suggest that machine learning models, especially tree-based regressors like Extra Trees, X-Gradient Boost, and Random Forest, can effectively predict medical insurance costs with high accuracy. By leveraging predictive modeling, both insurance providers and customers can make informed decisions regarding cost estimation and plan selection.