Predicting passenger loading level on a train car: A Bayesian approach
For transit agencies and passengers, this work provides a method to predict crowding, potentially improving service quality and operational efficiency.
The paper addresses the problem of predicting passenger loading levels on train cars to mitigate overcrowding. It proposes a Bayesian approach that achieves accurate predictions, enabling better crowd management.
Crowding in train cars is increasingly a major concern for transit agencies. From the perspective of the passengers and the transit agencies, overcrowding of the train cars has several negative consequences such as: (i) extended duration of passengers boarding and alighting which leads to longer dwell times, (ii) subsequent disruption of the headway and the schedule, and (iii) passenger dissatisfaction (e.g. increased stress and lack of privacy). Moreover, overcrowding during peak service hours also indicates inadequate infrastructure to meet the passenger demands. Realizing the importance of the crowding issue, transit agencies have developed measures to assess the crowding levels. The Transit Capacity and Quality of Service Manual provides guidelines on thresholds for crowding in transit systems in the United States.