Data-driven geofencing approach to identify bus route segments for travel time improvements using automatic vehicle location (AVL) data

Public transport (PT) systems frequently face schedule deviations and operational disruptions, which demand proper analytical frameworks to formulate effective interventions. This study develops an automated geofencing algorithm for PT route segmentation and classification using large-scale AVL data. It reduces the dimensionality of PT operational performance analytics and allows the identification of the critical route sections for travel time improvements. The method detects sections where travel time acceleration measures are most needed by applying statistical analysis and unsupervised machine learning. The approach offers a practical, evidence-based decision support tool for PT agencies aiming to enhance reliability and efficiency.