Anticipating Passenger Flow amid Pandemic-Induced Lockdowns: A Forecasting Approach

During the pandemic, the urban subway system underwent substantial changes in passenger flow, revealing distinct patterns. Fears of infection prompted a widespread reduction in public transportation use, particularly the subway. Many individuals turned to alternatives such as cycling, walking, and private vehicles. Concurrently, several subway stations were temporarily shuttered to contain the virus, leading to a marked drop in passenger numbers, especially during peak hours and in commercial districts. However, even with operational stations, pre-pandemic passenger levels remained elusive. Despite a gradual easing of restrictions as the pandemic subsided, the resurgence of subway ridership was not as swift as anticipated due to persistent shifts in travel behavior and lingering safety concerns. Consequently, subway operators and urban planners need to swiftly adopt adaptable measures. To address this dynamic and prolonged passenger flow pattern, a novel approach is introduced in this study: a pandemic-controlled station passenger flow prediction model. This model aims to comprehensively understand the spatiotemporal relationships among stations and their socioeconomic effects. By employing a random forest model to capture intricate spatiotemporal characteristics, the model accurately predicts passenger flow changes during control periods. This not only facilitates an understanding of evolving passenger dynamics but also offers vital insights for enhancing subway operations and strategic planning.