The Comparison of Forecasting Models for Individual Rider Insurance Premiums: Health Insurance Type

Abstract

          This research aimed to 1) forecast the individual rider insurance premiums: health insurance type using Box–Jenkins method: ARIMA Model and Multiple Linear Regression method 2) to compare the forecasting methods for health insurance premiums, and examine the accuracy of forecasting methods via the criteria of Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE). The data were the individual rider insurance premiums: health insurance type from the website of The Thai Life Assurance Association during January 2020 to September 2024, totally 57 months; a forecasting model, from October 2024 to September 2025, totally 12 months, was constructed to test and compare its accuracy.  This study was quantitative research utilizing R Studio software for data analysis by the Box–Jenkins method: ARIMA Model and multiple linear regression method.

           The results of the research found that following.

          1. Box–Jenkins method: ARIMA Model gave accuracy at 98.27% RMSE of 85.37% MAE of 63.77% and MAPE of 1.73% and Multiple Linear Regression method gave accuracy at 91.80% RMSE of 93.65% MAE of 89.32% and MAPE of 8.29%.

          2. Box–Jenkins method: ARIMA Model was the appropriate method and had the highest forecasting accuracy of the minimum RMSE and MAPE was at 1.73% for the health insurance premiums in 12 months. Therefore, it was accurate and appropriate for this purpose.