Time Series Analyses of UK Outbound Travel by Air

Abstract

Time series models are applied to outbound travel flows by air from the United Kingdom to 20 destinations. Objective assessments are made as to whether seasonal components in ARIMA processes should be modeled additively or multiplicatively. A selection strategy is presented to differentiate between the difference filters applicable to additive and multiplicative ARIMA models. The strategy inherently incorporates the possibility of increasing stochastic or deterministic seasonal variation. It is shown that neither logarithmic transformations nor a large number of difference filters are required. Selected ARIMA models tend to outperform Holt-Winters and Naïve models in terms of goodness of fit and forecasting accuracy.