Air transportation demand forecast through Bagging Holt Winters methods

This paper expands the fields of application of combined Bootstrap aggregating (Bagging) and Holt Winters methods to the air transportation industry, a novelty in literature, in order to obtain more accurate demand forecasts. The methodology involves decomposing the time series into three adding components: trend, seasonal and remainder. New series are generated by resampling the Remainder component and adding back the trend and seasonal ones. The Holt Winters method is used to modelling each time series and the final forecast is obtained by aggregating the forecasts set. The approach is tested using data series from 14 countries and the results are compared with five methodology benchmarks (SARIMA, Holt Winters, ETS, Bagged.BLD.MBB.ETS and Seasonal Naive) using Symmetric Mean Absolute Percentage Error (sMAPE). The empirical results obtained with Bagging Holt Winters methods consistently outperform the benchmarks by providing forecasts that are more accurate.