Study of tourism flow forecasting based on a seasonally adjusted particle swarm optimization-support vector regression model

Forecasting tourism flow accurately is one of the most important priorities for tourism destinations. Tourism flow has a distinct seasonal trend. Using a seasonal adjustment method to preprocess the sample data can eliminate seasonal influence, and can improve the accuracy of tourism flow forecasting. Support vector regression is a good method for machine learning, and is very suitable for predicting tourism flow. Supplemented by the use of the particle swarm optimization algorithm to choose appropriate parameters, SVR can get prediction results more accurately. Thus, we construct a PSO-SVR tourism flow forecasting model which takes seasonal influences into account. Sanya (Hainan, China) is taken as an example for empirical research. The fitting degree of different prediction models indicates that the predictive accuracy of the seasonally adjusted PSO-SVR model is significantly higher than that of the SVR model, the seasonally adjusted SVR model and the PSO-SVR model. That is to say, the seasonally adjusted PSO-SVR model is the most effective tool for predicting tourism flow. Copyright © 2015 Binary Information Press.