Scholarlink

The paper aims to enhance the risk management procedure in health insurance using an adaptive machine learning approach. The study starts with the data preprocessing activity that involves checking and/or dealing with missing values, converting nominal to quantitative variables, and scaling the features. The data is then divided into the training set and the testing set to have effective measures for model evaluation. The given preprocessed data is used to train each different machine learning model, and then its efficacy is measured with the help of Mean Absolute Error (MAE), Mean Squared Error (MSE), and percentage R-squared (R2) score. Comparing the various models that were trained, ensemble methods such as the Random Forest and XG Boost showed better results in predictive accuracy. The model of Random Forest disclosed the smallest MAE of 2545. 21 and the highest value of R2 of 0. A score of 91 shows a good performance of the model in predicting the insurance charges. Same way, XGBoost was also good with an Moreover MAE of 2765. 75 and an R2 score of 0.91. However, in the current study, the SVM model presented a very low R2 value turning to be negative making it less applicable in the current dataset. Finally, the author opines that Ensemble methods were effective in the insurance charges prediction and managing the health insurance risk, especially the Random Forest and XG Boost. These models could be embedded into health insurance systems to improve risk management solutions, hence implying accurate means of premium pricing and low risks.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing Health Insurance Risk Management with Ensemble Machine Learning Models: A Comparative Study of Predictive Accuracy

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Enshirah Ibrahim Altarawneh,
  • Khalid Alemerien

摘要

The paper aims to enhance the risk management procedure in health insurance using an adaptive machine learning approach. The study starts with the data preprocessing activity that involves checking and/or dealing with missing values, converting nominal to quantitative variables, and scaling the features. The data is then divided into the training set and the testing set to have effective measures for model evaluation. The given preprocessed data is used to train each different machine learning model, and then its efficacy is measured with the help of Mean Absolute Error (MAE), Mean Squared Error (MSE), and percentage R-squared (R2) score. Comparing the various models that were trained, ensemble methods such as the Random Forest and XG Boost showed better results in predictive accuracy. The model of Random Forest disclosed the smallest MAE of 2545. 21 and the highest value of R2 of 0. A score of 91 shows a good performance of the model in predicting the insurance charges. Same way, XGBoost was also good with an Moreover MAE of 2765. 75 and an R2 score of 0.91. However, in the current study, the SVM model presented a very low R2 value turning to be negative making it less applicable in the current dataset. Finally, the author opines that Ensemble methods were effective in the insurance charges prediction and managing the health insurance risk, especially the Random Forest and XG Boost. These models could be embedded into health insurance systems to improve risk management solutions, hence implying accurate means of premium pricing and low risks.