Forecasting short-term freight transportation demand - Poisson STARMA model

A framework for analyzing, describing, and forecasting freight flows for operational and tactical purposes is presented. A dynamic econometric model is proposed. This model incorporates the spatial and temporal characteristics of freight demand within a stochastic framework. The model was applied in an actual context and its performance was compared with standard time series models (benchmark) for forecasting ability. The proposed model outperformed the benchmark from the econometric viewpoint. Extensive diagnostic checking and sensitivity-analysis confirmed the robustness of the modeling methodology for short-term forecasting applications.