Hybridizing GMDH and least squares SVM support vector machine for forecasting tourism demand

Abstract

In this paper, we proposed a novel hybrid group method of data handling least squares support vector machine (GLSSVM) algorithm, which combines the theory a group method of data handling (GMDH) with the least squares support vector machine (LSSVM). With the GMDH is used to determine the inputs of LSSVM method and the LSSVM model which works as time series forecasting. The aim of this study is to examine the feasibility of the hybrid model in tourism demand forecasting by comparing it with GMDH and LSSVM model. The tourist arrivals to Johor Malaysia during 1970 to 2008 were employed as the data set. The comparison of modeling results demonstrate that the hybrid model outperforms than two other nonlinear approaches GMDH and LSSVM models.