Research on intercity travel mode recognition and network structure characteristics based on complex network and...

Area of Science:

• Transportation Science and Systems
• Network Analysis
• Data Mining and Machine Learning

Background:

• Identifying diverse intercity travel modes in high-density areas is challenging.
• Understanding spatio-temporal variations in intercity travel networks is crucial for efficient transportation planning.
• Existing methods may lack precision in mode identification and network analysis.

Purpose of the Study:

• To develop and validate a novel management model (CNT-RFC) for identifying intercity travel modes.
• To analyze the network structure characteristics of intercity travel over different time periods.
• To uncover the spatio-temporal heterogeneity of intercity travel patterns.

Main Methods:

• Integration of Complex Network Theory (CNT) with a Random Forest Classification (RFC) algorithm.
• Utilization of publicly available migration and transportation data (Jan 2021-Dec 2023).
• Extraction of network features (degree distributions, centrality, community detection) and application of RFC for mode identification.

Main Results:

• The CNT-RFC model achieved high accuracy (0.947), precision (0.928), and F1 score (0.947) for leisure travel identification, outperforming advanced models.
• Network analysis showed significant structural changes during holidays (e.g., decreased small-world coefficient, increased travel distance).
• Sensitivity analysis confirmed model robustness during the pandemic, highlighting asymmetric impacts on transportation networks.

Conclusions:

• The CNT-RFC model offers a precise and robust solution for identifying intercity travel modes and analyzing network dynamics.
• Findings provide critical insights into spatio-temporal travel pattern heterogeneity, essential for regional transportation planning.
• The study supports evidence-based policy formulation to address congestion and optimize intercity transportation efficiency.