Key points are not available for this paper at this time.
In operations of road freight transport, demand forecasts are mostly used for downstream optimization, following a predict then optimize setting. While the literature provides numerous optimization models that rely on accurate forecasts to support the planning of road freight transport, dedicated work on forecasting is limited, hindering a holistic treatment of the problem. Moreover, there is a disconnect to advances in the broader forecasting literature, limiting the adoption of modeling innovations and methodological advances. These can harm the quality and validity of forecasts designed to support road freight transport. We link the relevant forecasting publications to different prominent optimization problems for road freight transportation, highlighting disconnects between forecasting and optimization models in the area. By contrasting these with the current discourse in the forecasting literature, we identify relevant modeling and methodological improvements, in model building, the choice of loss function, and evaluation. These are important to better link forecasts with optimization models for road freight transport. Furthermore, we propose a unified hierarchical framework for freight demand to support aligned decisions across planning levels, which is an important consideration for practice. Our review helps structure and steer academic research and provides opportunities to improve existing forecasting models; a closer integration with optimization; and ultimately improve decisions. • Holistical literature review on road freight demand. • Connects road freight demand forecasting with optimization. • Shows how to improve the interface between forecasting and optimization. • Provides a hierarchical framework for road freight demand forecasting and operations.
Sonnleitner et al. (Wed,) studied this question.