Optimization of Intercity Multimodal Network Considering Different Travel Behaviors Estimated via Non-negative Matrix Factorization

抄録

Efficient intercity travel networks are important for economic growth and environmental sustainability, yet traditional networks often overlook the influence of travel behaviors. Therefore, in this study, we address the knowledge gap by incorporating travel behaviors into intercity transportation networks in Japan using non-negative matrix factorization (NMF) on mobile-phone location data. The NMF analysis revealed three key travel-behavior patterns—Business, Personal reasons, and Sightseeing/Leisure, whose seasonal dynamics were examined. We modeled these variations to develop an optimized multimodal transportation network integrating rail and air services to minimize social costs. The findings show that incorporation of travel behaviors, i.e., the necessity of hybrid transportation networks, particularly for high travel volumes and significant travel-time differences, results in drastic changes in the optimal network shape. In addition, the relevance of single-modal networks under specific conditions is highlighted, demonstrating the potential of NMF for identifying travel behaviors and improving transportation network designs.