Journal of Statistics Applications & Probability


This paper aims to use a hybrid ARIMA-ANN model for time series forecasting by combining Auto Regressive Integrated Moving Average (ARIMA) model and the Artificial Neural Networks (ANN). The hybrid ARIMA-ANN model is flexible enough to capture two kinds of time series: the linear model that can only model the linear relationship and nonlinear that can only model the nonlinear relationship. The time series data for the Crude Oil (petroleum) Monthly Price - Saudi Riyal per Barrel was used during the period from Jul-2001 to May-2021, which represents 239 observations. The first 215 observations are used as train series and the last 24 observations are used as testing series. The accuracy measures, Mean Square Error (MSE), Mean Absolute Percentage Error (MAPE), and Mean Absolute Error (MAE) for the hybrid combination ARIMA and ANN were compared against ARIMA and ANN methods. The results indicate significant improvement in (MSE, MAPE, and MAE) values for the hybrid ARIMA-ANN method.

Digital Object Identifier (DOI)