Tourism Demand Forecasting for Turkey by Using Adaboost Algorithm

Tourism is associated with rest and vacation for people who spend most of their day working in order to keep up with the developing and changing world standards and to live a decent life. With the increasing demand for tourism in recent years, the place of tourism in the world economy continues to rise day by day. Considering its contribution to the economy, especially in developing countries, it is very important to know the tourism demand forecast in advance. Turkey is one of the most visited and rising trend countries in the tourism sector. In this study, tourism demand forecast for Turkey in general is made with three different machine learning based methods, k-nearest neighbour, decision tree, and AdaBoost. A new data set was created by determining the features that may affect the tourism demand. Accordingly, cross exchange rate in dollars, consumer price index, tourism expenses and month are considered as independent variables, and the number of visitors is considered as dependent variable. Experimental results show that all three methods yielded successful results, but the AdaBoost method showed the highest R
²
(0.88315) value. As a result, it has been observed that the machine learning-based methods used in the study are effective and efficient tools in predicting tourism demand.