A machine learning approach for the operationalization of latent classes in a discrete shipment size choice model. (January 2019)

• Record Type: Journal Article
• Authors: Piendl, Raphael; Matteis, Tilman; Liedtke, Gernot
• Is Part Of: Transportation research. Volume 121(2019)
• Page Start: 149
• Page End: 161
• Publication Date: 2019-01
• Subjects: Freight transport -- Shipment size -- Latent class analysis -- Machine learning -- Bayesian classification; Logistics -- Periodicals; Transportation -- Periodicals; 388.011
• DOI: 10.1016/j.tre.2018.03.005
• Languages: English
• ISSNs: 1366-5545

Abstract

Highlights: Introduction of machine learning techniques to freight transport modeling. Bayesian classifier used for large-scale operationalization of latent segments. Moderate and reasonable sensitiveness to a measure which induces tariff variations. Behavioral heterogeneity captured by latent segments based on logistics requirements. Abstract: This paper elaborates a novel approach for implementation of latent segments concerning behaviorally sensitive shipment size choice in strategic interregional freight transport models. Discrete shipment size choice models are estimated for different homogenous segments formed by latent class analysis. A machine learning technique called Bayesian classifier is applied to link segments obtained from a sample to data of commodity flows being available on a national level. Finally, in an exemplary scenario, the impact of information and communication technologies on shipment size distributions is calculated, revealing moderate elasticities and a predominant substitution of less than truck loads by full truck loads.