Developing a Composite Model (Multiple Linear Regression and ARIMA Models) to Forecast Malaysian Imports

Abstract

Forecast models have played a major role in many statistical applications for more than a century. When the error term follows a normal distribution, these models can yield highly accurate prediction results. The appropriate methodological framework for analysing time series data provides the theoretical basis for applying such models. This study, therefore, seeks to identify an appropriate statistical model that can better predict Malaysia’s imports by evaluating several approaches, including the Autoregressive Distributed Lag (ARDL), the Autoregressive Integrated Moving Average (ARIMA), and a composite (a combined regression–ARIMA) model proposed in this research. The proposed model integrates insights from both regression and ARIMA approaches, thereby facilitating the development of a more effective statistical framework for forecasting the volume of Malaysia’s imports and enhancing existing prediction methods. The prediction performance is assessed using the Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and the coefficient of determination (R²). The results show that the combined model outperforms all other models in terms of accuracy. Its primary strengths lie in its predictive power and its ability to address regression model issues such as residual autocorrelation. Model parameters and predictions were estimated using 54 observations. Future research may expand on these findings by exploring other approaches, such as combined models that account for autocorrelation or heterogeneity problems, or by applying larger datasets on Malaysia’s imports and comparing the results with those of the present study.