An efficient multi-modal urban transportation network partitioning approach for three-dimensional macroscopic fundamental diagram

The three-dimensional macroscopic fundamental diagram (3D-MFD) provides a comprehensive understanding of the relationship between network efficiency and accumulation of cars and buses in transportation networks. In this paper, we propose a novel approach to partition multi-modal urban transportation networks into several homogeneous sub-regions to obtain well-shaped 3D-MFDs for each sub-region. The proposed approach consists of three stages: an initial partitioning process using the Symmetric Non-negative Matrix Factorization (SNMF) and Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS), iterative merging of the initial partitioned sub-regions based on SNMF and TOPSIS, and optimization of the sub-regional boundaries, considering the traffic network's link hierarchy, to accurately capture its physical characteristics. Comparative experiments utilizing real data from the Zurich traffic network demonstrate that our proposed method achieves remarkable partitioning results, as supported by evaluation metrics. Furthermore, the partitioning results obtained through our proposed approach exhibit more favorable network physical characteristics, providing a solid foundation for implementing control strategies to enhance network efficiency.