A Hierarchical Passenger Mobility Prediction Model Applicable to Large Crowding Events

Predicting individual mobility of subway passengers in large crowding events is crucial for safety management and crowd control. However, most previous models focused on prediction under ordinary conditions. Here, we develop a passenger model, which also applicable to events. The developed model includes the trip-making part trip attribute part. For prediction, regularized logistic regression that employs proposed cumulative features, number potential trips, generation index. an n -gram incorporating new feature, attraction index, each cluster passengers. incorporation three features clustering considerably improves accuracy especially