Forecasting Philippines imports and exports using Bayesian artificial neural network and autoregressive integrated moving average

Abstract

In this research, Autoregressive Integrated Moving Average (ARIMA) and Bayesian Artificial Neural Network (BANN) were used in forecasting the imports and exports of the Philippines and the comparison of two models are one of the main objective of this research. The data were gathered from Philippines Statistical Authority with a total of 100 observations from the first quarter of 1993 to fourth quarter of 2017. Furthermore, it can be determined in this research the best fit among the models in forecasting the imports and exports of the Philippines and the researchers will give the forecast values of imports and exports from the first quarter of year 2018 to the fourth quarter of year 2022 using the most fitted model. The researchers conducted a Statistical test in order to formulate and compare the statistical models of ARIMA and BANN for imports and exports then applied the forecasting accuracy such as MSE, NMSE, MAE, RMSE, and MAPE to compare the performance of the two models. By comparing the results, the researchers concluded that Bayesian Artificial Neural Network is the most fitted model in forecasting the imports and export of the Philippines. Upon using the Paired T-test, the p-value for both imports and exports are greater than the level of significance (α = 0.01) which means that there is no significant difference between actual and predicted values for both imports and exports of the Philippines. This study could help the economy of the Philippines by considering the forecasted Imports and Exports which can be used in analyzing the economy's trade deficit.