Forecasting Foreign Guest Nights in Colombo and Greater Colombo of Sri Lanka

The forecasting occupancy guest nights is an essential discipline in Marketing Strategy, Yielding across various channels, purchase decisions, expansion plans and various decisions about staffing and hotel maintenance in tourist hotels. Colombo city and Greater Colombo is the highest regions of night occupancy by international tourist in Sri Lanka. The high occupancy will increase the demand for accommodation. Therefore the hotel industry should adopt revenue management practices to maximize profits and optimize operations. This study was focused on identifying suitable forecasting techniques for occupancy guest nights of international tourist. Monthly data of foreign guest nights for the period of January 2008 to December 2016 were obtained from Sri Lanka Tourism Development Authority (SLTDA). The Decomposition additive and multiplicative models and Seasonal Auto-regressive Integrated Moving Average (SARIMA) were tested for forecasting. The Anderson–Darling test, Auto-Correlation Function (ACF), Durbin-Watson test, and Ljung-Box Q (LBQ)-test were used as the goodness of fit tests in model validation. The best fitting model was selected by comparing the relative measurements of errors and the absolute measurements of errors. The residuals of both Decomposition; additive and multiplicative models were found normally distributed, but not independent. Both relative and absolute measurements of errors of the model ARIMA (1,0,1)(2,1,1)4 are very low in model fitting. The residuals of ARIMA (1,0,1)(2,1,1)4 were found normally distributed and independent. Therefore, future night occupancy by the foreign guest in Colombo city and Greater Colombo can be forecast ed by past night occupancy by foreign guest, past errors and seasonal components. The study concluded that SARIMA performs better than Decomposition models in forecasting occupancy guest nights. However, the SARIMA model is not capable in capturing the cyclical variations. Therefore, it is recommended to test the Circular Model for de-trended data for better forecasting