Forecasting monthly international air passenger traffic at the Entebbe International Airport, Uganda

Abstract

This study thought to forecast the number of international air passengers at Entebbe International Air Port (EIA) to aid management in making decisions. The study utilized secondary data from Uganda Civil Aviation Authority (UCAA) of monthly total international air passenger flows from January 1990 to December 2017 and three forecasting approaches were compared. The conventional linear predictive models of Seasonal Autoregressive Moving Average (SARIMA), Holt Winters Exponential Smoothing (HWES) and the nonlinear time series model, the Self Exciting Threshold Autoregressive (SETAR). The findings indicated that the series, international air passenger flow series at EIA exhibited both trend and seasonality and followed a nonlinear trend which behavior is better modeled by a nonlinear threshold model. The study fit the two-regime nonlinear SETAR model and compared its in-sample and post-sample performance with that of the linear SARIMA and HWES models. The in-sample/ goodness of fit performance of three models assessed using the AIC and MAPE at 95% confidence interval suggested that the nonlinear SETAR model outperformed the linear SARIMA and HWES models. However, for the post-sample performance, the time series plot and the RMSE suggested that the linear HWES (RMSE=3095) model out performs the SARIMA (RMSE=3253) and the SETAR (RMSE=16848) models and therefore should be used to predict future passenger flows. The study, thus, recommends use of HWES model to predict air passenger traffic at Entebbe International Airport.