Deep Learning Framework for Forecasting Tourism Demand

Abstract

Accurate Tourism demand forecasting plays an important role to make decision and plan policy. However, tourism demand is characterized by complexity and non-linearity. Traditional tourism demand forecasting techniques are Linear methods and unable to fully simulate the nonlinear characteristics of tourism demand. Deep learning (DL) methods can be a promising solution to achieve an accurate forecast. These models are able to evaluate the non-linear relationship, without the drawbacks of Time Series and econometric models. In this paper, a deep learning Models are proposed to accurately predict tourist arrivals for Morocco from 2010 to 2019. The proposed framework uses a long short-term memory (LSTM) and gated recurrent unit (GRU). Experiments demonstrate that the LSTM and GRU methods perform better than support vector regression (SVR) and artificial neural network models (ANN).