Forecasting with Fourier Residual Modified Arima Model: The Case of Air Cargo in Taiwan

Purpose - Import-export activities normally require the involvement of different sectors needing proper plans to make the trade-flow stably grown. This paper is aimed at establishing an accurate forecasting model for trade volume. Design/methodology/approach- In this study, a new forecasting model named ARIMAF/ SARIMAF was proposed by combining the Fourier series with the conventional ARIMA/ SARIMA forecasting model. Findings- In the cases of historical data of the volumes of imported and exported air cargo in Taiwan, SARIMAF(3,1,1)(1,1,1)12 and SARIMAF(2,1,3)(1,1,1)12 models were found fitting well with the mean absolute percentage error values of 0.0104 and 0.0096, respectively. Using these two models, the monthly volumes of air cargo in Taiwan in 2012 are predicted. Originality/value- The using Fourier series to modify the residuals of the traditional ARIMA is first proposed in this paper.