Forecasting Türkiye's International Tourism Demand

Abstract

The structure of tourist demand is sensitive since it is very easily affected by the consequences of economic, political, and social crises. Since the limited ability to increase tourism supply, it is crucial to analyze the demand structure and develop suitable strategies. This outcome can only be achieved by an accurate and effective demand prediction. There is no singular approach that ensures success in demand forecasting. Hence, in order to estimate demand accurately, it is advisable to create many models and choose the one with the lowest error rate. This study aimed to develop the best-performing prediction model by using monthly data of international visitors who visited Türkiye from January 2002 to August 2023 and stayed in Tourism Ministry-certified accommodations. Within this framework, the data was first analyzed to identify the trend and seasonal component. Afterwards, various models were employed including Naive III, simple moving average, double moving average, seasonal exponential smoothing, and artificial neural networks. The data generated by these models has been analyzed by comparing it with the actual data from the last 24 months, using MAPE and RMSE results. According to the research findings, it has been determined that artificial neural networks produce the most accurate results.