The Comparison of Forecasting Models for Total Premiums of Life Insurance Companies in Thailand

Abstract

The purpose of this research, in the study of total premiums of monthly life insurance companies in Thailand during a period of 72 months from January 2017 to December 2022, was to study and compare the forecasting performance of four forecasting models, 1) Decision Tree Regression model, 2) Random Forest Regression model, 3) Support Vector Regression (SVR) model and 4) Polynomial Regression model. The forecasting efficiency was compared in the four models using Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE). The results of the research showed that 1) the Decision Tree Regression model with 1654.00 of RMSE and 2.93% of MAPE, 2) SVR model with 5560.59 of RMSE and 9.03% of MAPE, 3) Polynomial Regression model with 6283.63 of RMSE and 11.36% of MAPE and 4) Random Forest Regression model with 6723.48 of RMSE and 11.76% of MAPE. The model with the suitable forecast performance is the Decision Tree Regression, but the Decision Tree Regression model is not suitable for long-term forecasting, so then the SVR model was chosen as a better fit for long-term forecasting.