Abstract
ABSTRACT Discovering metro passenger flow recovery patterns from historical unplanned disruptions enables operators to better prepare for a new disruption. The task is challenging as passenger flow recovery patterns are generated by different types of disruptions at different locations and times. We start with the Robust Principal Component Analysis (RPCA), a dimensionality reduction technique to calculate passenger flow change under disruption and quantify disruption impact. With this information, we modify the Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) model with the tailor-made discrete input module and distance metric module, to discover the typical passenger flow recovery patterns with different shapes and sizes. Our approach is validated in the Hong Kong Mass Transit Railway system. The results show that our approach can discover interpretable recovery patterns that are easily neglected by other clustering methods, because of its ability to quantify and distinguish disruption impact on normal passenger flow.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call