Application of Artificial Neural Network in Forecasting Tourist Demand to Quang Binh, Vietnam

Abstract

Accurate forecasting of tourist demand is crucial for guiding policies, planning, and developing strategies for a locality or country. There are various approaches to forecasting tourist demand, among which time series data-based forecasting attracts the most attention due to the unstructured nature of this type of data. Artificial neural network is evaluated as a particularly suitable forecasting method for unstructured data, although it is almost impossible to explain its internal processes. This paper uses an artificial neural network to forecast time series data on tourist demand to Quang Binh. Three network models, MLP (Multi-Layer Perceptron), RBF (Radial Basis Function), and ELN (Elman network), are evaluated. With the obtained simulation results, the RBF network provides the best forecasting performance, with the lowest MSE (Mean Squared Error), RMSE (Root Mean Square Error), MAE (Mean Absolute Error), and MAPE (Mean Absolute Percentage Error), compared to the other two network types. This result further confirms that the feature of transforming non-linear space into linear space of the hidden layer has made RBF powerful for unstructured data.