Quantifying volatility characteristics of passenger flow in metro stations based on rolling-window

Understanding the characteristics of passenger flows within a metro network is essential to enhance the operational efficiency and safety of urban rail transit system. Most research has focused on investigating trends or patterns of passenger flows, ignoring the volatility inherent in these trends or patterns. This study employs a rolling-window analysis approach, coupled with two nonparametric methods and on statistical method to quantify the volatility in passenger flow patterns across metro stations. The volatility characteristics of passenger flow are captured by the ranges of potential changes in passenger flows, which are measured using three different indices: kickoff, width, and width-to-flow ratio. By employing actual passenger data from six metro stations in Suzhou, China, this study quantifies the volatility characteristics of passenger flows on weekdays and weekends. It reveals that passenger flow volatility on weekends is greater than on weekdays, indicating more erratic passenger flow changes on weekends. Our findings have practical implications for enhancing management's understanding of passenger flow volatility and the provision of reliable metro information to travelers via information systems. Future work is recommended to develop and test more methods for quantifying the volatility of passenger flow in the metro stations, metro lines, and the metro network.