Forecasting Foreign Visitors Arrivals Using Hybrid Model and Monte Carlo Simulation

The tourism industry is one of the important revenue sectors in today’s world. Millions of visits are made monthly to different countries across the planet. Some countries host more tourists than others, depending on the availability of factors that would fascinate visitors. Tourism demand can be affected by different factors, which may include government policies, insecurity, political motive, etc. Being an important sector, policymakers/governments are keen on models that would provide an insight into the inherent dynamics of tourism in their country. Especially in forecasting future tourist arrivals, as it will greatly assist in decision making. Several tourism demand models have been presented in the literature. The best practice is to have a model that would account for uncertainty in estimations. In this paper, an ANN-Polynomial-Fourier series model is implemented to capture and forecast tourist data for Turkey, Japan, Malaysia, and Singapore. The proposed model is a combination of the artificial neural network (ANN), polynomial fitting (poly), and Fourier series fitting (Fourier). The proposed model is designed to capture the data trend component using the polynomial fitting, the data seasonal component using the Fourier series fitting, and other data anomalies using the artificial neural network. Multistep ahead forecasting is made for each of the studied tourist data, and estimation uncertainties are covered by generating multiple forecast paths (Monte Carlo forecast). According to estimations, Turkey will expect a 10.22% increase in 2021 compared to the tourist arrivals it received in 2020. Japan is expected to have a 92.42% decrease in 2021 compared to the tourist arrivals it received in 2020. Malaysia is also expected to have a 54.81% decrease in 2021 when compared to the number of tourists it received in 2020. Finally, Singapore will expect a 70.55% decrease in 2021 compared to the number of tourists it received in 2020.