This research explores the use of machine learning classifiers for modeling freight mode choice. Eight common machine learning classifiers and the Multinomial Logit model are investigated using the 2012 Commodity Flow Survey data. The performance of these classifiers is compared based on their prediction accuracy. The study finds that tree-based ensemble classifiers, particularly Random Forest, Boosting and Bagging, provide the most accurate predictions. Shipment characteristics like distance, industry classification of the shipper and shipment size are identified as the most significant factors for freight mode choice decisions.
Publication date: 2 Feb 2024
Project Page: Not provided
Paper: https://arxiv.org/pdf/2402.00659
