Performance Comparison of Holt-Winters and SARIMA Models for Tourism Forecasting in Turkey

Forecasting the number of tourists coming to Turkey can play a vital role in strategic planning for both private and public sectors. In this study, monthly data of foreigners visiting Turkey were collected between the years 2007 and 2018. The data showed a seasonal behavior with an increasing trend; consequently, two methods were chosen for the study: Holt-Winters (HW) and Seasonal Autoregressive Integrated Moving Average (SARIMA). The objective of the study is to determine the most appropriate forecasting model to achieve a good level of forecasting accuracy. The findings showed that all models provided accurate forecast values according to error measures. However, multiplicative model of HW achieved the highest forecasting accuracy followed by SARIMA and additive HW respectively.