Time Series Analysis for Modeling and Forecasting International Tourist Arrivals in Sri Lanka

Abstract

Tourism is one of the income generating industries in a developing country which directly contribute to the economy. Therefore, forecasting tourist arrivals is important for making policy decisions to improve facilities and other related factors in this industry. In this paper, an attempt has been made to forecast tourists’ arrivals using time series modelling. The time span used for the study is from January 2000 to February 2016. In the modelling exercise, data has been analyzed based on the two sets of data; long-term (2000-2016), and post-war (2010-2016). This categorization was due to the significant change in the industry after the end of the civil war in 2009 in Sri Lanka. An Auto-Regressive Integrated Moving Average (ARIMA) method and Multiplicative Decomposition approach (MDA) were employed to model the data. When the forecasts from these models were validated, post-war data has more accurate results having low Mean absolute percentage error (MAPE) for MDA than the ARIMA approach. The comparison of actual data with the predicted values also confirmed that the MDA model obtained from the post-war series has high predictive ability.