Abstract:
In this research study, the approach of Holt- Winter’s Method and Seasonal
Autoregressive Integrated Moving Average (𝑆𝐴𝑅𝐼𝑀𝐴) method were implemented to forecast
tourist arrivals in Sri Lanka. In this case, Sri Lankan monthly tourist arrivals data from January
2000 to December 2017 was considered. In the modelling implementation, data was analysed
based on the two types of data such as long-term (2000-2017) and post-war (2010-2017).
Because of the Sri Lankan Civil War ended in 2009, the data were categorized into two types.
After the Sri Lankan civil war, tourist arrivals have increased annually. For that, forecasting Sri
Lankan tourist arrivals is a necessary topic to build policy resolutions to enlarge conveniences
plus additional interconnected issues in this industry. The first order difference data was
concerned to make the data as stationary for the ARIMA approach. The best Holt- Winter’s
model was selected based on the least Root Mean Square Error (RMSE) and Mean Absolute
Deviation (MAD) values meanwhile the best 𝑆𝐴𝑅𝐼𝑀𝐴 model was selected based on the
minimum Akaike Information Criterion (𝐴𝐼𝐶) value. The required statistical analysis was
performed using Solver tool in Excel, Eviews9 and Minitab-16 software at 5% of significance
level. The results reveal that for the long-term and post-war period, 𝐴𝑅𝐼𝑀𝐴 (3, 1, 2) (1, 0, 1)12
and 𝐴𝑅𝐼𝑀𝐴 (2, 1, 3) (1, 0, 0)12 are the suitable models respectively. Among the two approaches,
𝐴𝑅𝐼𝑀𝐴 (2, 1, 3) (1, 0, 0)12 for post-war is the best model to sketch and to forecast the monthly
tourist arrival pattern in Sri Lanka since having the least RMSE and MAD with a very precise
extent by it satisfies the model assumptions. As well as, it indicates that forecasted and actual
tourist arrivals are not much deviated from each other.
