Pattern detection in the vehicular activity of bus rapid transit systems

Area of Science:

• Transportation Science
• Network Science
• Urban Planning

Background:

• Bus rapid transit (BRT) systems are crucial for urban mobility.
• Understanding BRT operational dynamics is essential for service improvement.
• Existing methods may not fully capture system-wide and localized activity patterns.

Purpose of the Study:

• To explore novel methods for detecting activity patterns in bus rapid transit (BRT) systems.
• To analyze both infrastructure and vehicle movement aspects of transit operations.
• To characterize BRT systems using statistical and network science approaches.

Main Methods:

• Analysis of velocity and position records from nine BRT systems in the Americas.
• Statistical analysis of vehicle velocities at global and local scales.
• Application of Kullback-Leibler divergence for inter-zone activity comparison.
• Network representation of zone similarities and community detection algorithms.

Main Results:

• Identified collective patterns characterizing individual BRT systems.
• Detected distinct groups of zones with similar vehicle movement patterns using community detection.
• Demonstrated the utility of network representation for analyzing BRT activity at multiple scales.

Conclusions:

• Network science provides a powerful framework for characterizing BRT system activity.
• Geolocalized vehicle movement data can be effectively analyzed using network methods.
• The proposed approach is adaptable for analyzing other public transportation systems.