Forecasting Oil Price in Malaysia using ARIMA and Artificial Neural Network

Abstract

The importance and difficulty of accurately predicting future oil prices stem from the fact that these prices impact so many different economic and non-economic variables. Numerous factors, such as economic growth, political events, and individual ambitions, contribute to the high degree of unpredictability inherent in oil price forecasts. This study sought to forecast the future price of oil in Malaysia using the ARIMA and ANN models. The goal was to identify the model that would be most suitable for this study. Additionally, the purpose of the study was to make an estimate of the price of oil in Malaysia in the future. This study forecasts oil prices from September 2003 through June 2023 using the ARIMA and ANN models. The ARIMA (1,1,0) model was determined to be the most efficient based on its lowest p-value and MSE (Mean Square Error). Further analysis was conducted to assess the accuracy of predictions of the ARIMA and ANN  using three prediction accuracy metrics which are Mean Absolute Error (MAE), the Mean Absolute Percentage Error (MAPE), and the Root Mean Square Error (RMSE) . The results demonstrate that when comparing the two models, ANN outperforms the ARIMA (1,1,0) model in terms of forecast accuracy. The lower values of 36.30 for Mean Absolute Error (MAE), 14.23% for Mean Absolute Percentage Error (MAPE), and 48.81 for Root Mean Square Error (RMSE) show that the ANN is proficient. As a result of their higher precision in forecasting, artificial neural networks (ANN) are more suitable for their ability to anticipate the price of energy in Malaysia. Consequently, ANN were utilised in order to make predictions for the subsequent six months, which span from July 2023 to December 2023.