Evaluating the Performance of Linear and Nonlinear Models in Forecasting Tourist Occupancy in the Region of Western Greece

Abstract

Accurate tourism demand forecasting systems are very important in tourism planning, especially in high tourist countries and regions within. In this paper we investigate the problem of accurate tourism demand prediction using nonlinear regression techniques based on Artificial Neural Networks (ANN). The relative accuracy of the Multilayer Perceptron (MLP) and Support Vector regression (SVR) in tourist occupancy data is investigated and compared to simple Linear Regression (LR) models. The relative performance of the MLP and SVR models is also compared to each other. For this, the data collected for a period of 8 years (2005–2012) showing tourism occupancy of the hotels of the Western Region of Greece is used. Extensive experiments have shown that the SVM regressor with the RBF kernel (SVR-RBF) outperforms the other forecasting models when tested for a wide range of forecast horizon (1–24 months) presenting very small and stable prediction error compared to SVR-POLY, MLP, as well as the simple LR models.