This study aims to apply a new forecasting approach to improve predictions in the hospitality industry. In order to do so we develop a multivariate setting that allows incorporating the cross-correlations in the evolution of tourist arrivals from visitor markets to a specific destination in Neural Network models.