Forecasting Tourist Inflow in Bhutan using Seasonal ARIMA

The main aim of this paper is to generate one-period-ahead forecasts of international tourism demand for Bhutan by selecting appropriate model both ARIMA as well as exponential smoothing. Before selecting an appropriate model, formal stationary tests has been applied in this paper and finds that, the series are stationary at level. Secondly, in order to get a good estimation, this paper has identified the autoregressive (AR) and moving average (MA) of the entire period of the data. Therefore, the future demand of tourism is forecast based on the combination of AR and MA, which known as ARMA model. In this paper, the competing models have been thoroughly investigated when the model adequacy has been checked before the best combination of ARIMA model was selected. During the diagnostic stage we came to know that the data has a seasonal components thus, the best fitted ARIMA (0, 1, 1) (1, 1, 1) with seasonal effects or well known as SARIMA approaches has been suggested through this study and the forecasting process is based on this combination.