Forecasting Tourist Arrivals in China Based on Seasonal Decomposition and LSSVR Model

Abstract

In this study, a novel decomposition-ensemble methodology is proposed based on seasonal decomposition and the least squares support vector regression (LSSVR) model for forecasting foreign tourist arrivals in China. Firstly, the original time series of foreign tourist arrivals in China are decomposed into several components. Then, an individual forecasting model is selected for the prediction of a component. Finally, these prediction results of the components are combined as an aggregated output of forecasts of foreign tourist arrivals. Using the time series of tourist arrivals from twelve foreign countries to China, an empirical study is implemented for the purposes of illustration and verification. The results suggest that our proposed hybrid model TS-SL-LSSVR can achieve better forecasting performance than other models such as seasonal autoregressive integrated moving average (SARIMA), back propagation neural network (BPNN), and single LSSVR model.