Seasonal SVR with FOA algorithm for single-step and multi-step ahead forecasting in monthly inbound tourist flow

Accurate monthly inbound tourist flow forecasting can provide the reliable guidance for better tourism planning and administration. However, it has been found that the monthly inbound tourist flow demonstrates a complex nonlinear characteristic and an obvious seasonal tendency. Support vector regression (SVR) has been widely applied to handle nonlinear time series prediction, but it suffers from the key parameters selection and the influence of seasonal tendency. This paper proposes a novel approach, namely SFOASVR, which hybridizes SVR model with fruit fly optimization algorithm (FOA) and the seasonal index adjustment to forecast monthly tourist flow. Besides, in order to comprehensively evaluate the forecasting performance of the hybrid model, two kinds of forecasting horizons, namely single-step-ahead and multi-step-ahead, are used. In addition, the inbound tourist flow to mainland China from January 2000 to December 2013 is used as data set. The results show that the proposed hybrid SFOASVR approach is a viable option for tourist flow forecasting applications.