The Analysis for the Cargo Volume with Hybrid Discrete Wavelet Modeling

Abstract

Many efforts have been made to the development of models that able to analyze and predict marine cargo volume. However, improving forecasting especially marine cargo throughput time series forecasting accuracy is an important yet often difficult issue facing managers. In this study, a TEI@I methodology based hybrid forecasting model is proposed. The original time series are decomposed different scale components using discrete wavelet technique based on seasonality analysis of components. All decomposed components are predicted by radial basis function networks due to its flexible nonlinear modeling capability. Empirical results suggest that the use of discrete wavelet technique enhances the ability of monthly volatility mining and demonstrate consistent better performance of the proposed approach.