EMD-based model with cooperative training mechanism for tourism demand forecasting

Abstract

A novel EMD-based model with the cooperative training mechanism is proposed to enhance the tourism demand forecasting performance of tourist attraction. Our proposed model develops a cooperative training mechanism integrating a bi-level optimization model, a differential evolution algorithm and interactive fuzzy programming to optimize the neural networks of sub-series decomposed by EMD, which overcomes the difficulty of capturing the relationship among the sub-series and acquires the forecasting accuracy. To verify the performance of the model proposed in this study, extensive experiments are conducted. The experimental results illustrate the superiority of our model among all benchmark models and indicate its effectiveness in incorporating the cooperative training mechanism into the decomposition-ensemble model.