Tourism demand forecasting using a hybrid approach

Abstract

The goal of this study is to explore and evaluate the efficiency of a hybrid model ARIMA-SVR in tourism demand forecasting. The proposed model uses the data from Law and Au (1999) that includes 30 years of the number of Japanese visitors to Hong Kong. To illustrate the productivity of this model, we benchmark it with those two single models: ARIMA and SVR. The hybrid model is built based on using the residual results from the ARIMA model as input data for the SVR algorithm to train. Experimental results demonstrated that the forecasting efficiency of a hybrid model outperforms the two remaining models by 12.12% in MAPE. However, in terms of the RMSE index, the ARIMA model shows a better performance, which is 214357.50 compared to 229976.29. The hybrid algorithm's strength may assist the economy in general and the tourist sector in particular. The reliable projections are critical for tourist attractions where decisionmakers and company managers aim to capitalize on sector advances and/or balance their local environmental and economic performance.