Olympic Games passenger flow analysis and forecasting

Abstract

Using smart card data collected during the 19-day Paris Olympic Games, this study offers unique insights into the travel dynamics during this mega-event and the event’s impact on the public transport system during peak episodes. We identify six groups of departure flow profiles after game events that show distinct patterns in terms of the timing and volume of departure peak flows. A set of potential contributing factors including temporal, spatial, and event-related features are used to explain these clusters. The results show that difference among the departure profiles can be well characterized by event capacity, event duration, distance from the station to the stadium, distance from the station to Paris center, and event ending time. In addition, we applied forecasting models to predict passenger flow departure volumes at 30-minute intervals. Our results can support crowd management, optimize service capacity, and resources allocation effectively during large-scale events.