Short-Term Passenger Flow Prediction of Metro Stations around Sports Events Based on AFC Data

Activities lead to the concentration of people in the region, which not only causes great pressure on the surrounding infrastructure, but can also easily give rise to group events. This article collects the sports event information and AFC date for nearly 3 years and analyzes the traffic characteristics around the activities. Then it explores the date attribute, activity type, weather, home, and away teams which as the number of reported several influence factors on the influence of induced passenger flow at the station. Based on this, a gradient boosting decision tree (GBDT) prediction model was constructed and verified by an activity. The prediction results show that the average prediction accuracy of passenger flow out of Dongdaqiao, Dongsishitiao, and Tuanjiehu station in 15 min is 93.67%, 90.76%, and 89.61%. The prediction accuracy of inbound passenger flow are 80.68%, 78.96%, and 78.11%. This study can provide theoretical support for the formulation of metro emergency plans and passenger flow evacuation plans during large-scale activities.