Bayesian bootstrap aggregation for tourism demand forecasting

Abstract

Limited historical data are the primary cause of the failure of tourism forecasts. Bayesian bootstrap aggregation (BBagging) may offer a solution to this problem. This study is the first to apply BBagging to tourism demand forecasting. An analysis of annual and quarterly tourism demand for Hong Kong shows that BBagging can, in general, improve the forecasting accuracy of the econometric models obtained using the general-to-specific (GETS) approach by reducing, relative to the ordinary bagging method, the variability in the posterior distributions of the forecasts it generates.