Naïve Bayes-Based Transition Model for Short-Term Metro Passenger Flow Prediction under Planned Events
Document Type
Journal Article
Publication Date
2022
Subject Area
mode - subway/metro, place - asia, place - urban, ridership - demand, ridership - modelling, technology - passenger information, technology - ticketing systems
Keywords
metro, passenger demand, modelling, automatic fare collection
Abstract
Short-term passenger flow prediction under planned events is important to reduce passenger delay and ensure operational safety in metro systems. However, most studies make predictions under normal conditions. The study proposes a naïve Bayes transition model for short-term passenger flow prediction under planned events. The target prediction scenario identification is modeled as a binary classification problem using naïve Bayes. The sub-models are developed using gradient boosting decision tree (GBDT) and deep learning (DL) models for normal and planned event scenarios with predictor variables tailored to different passenger demand patterns. The sub-predictor from GBDT or DL is selected based on the inferred prediction scenario. The case study uses automatic fare collection (AFC) data of Shanghai and Hong Kong metro systems. The results show that the proposed model outperforms other representative individual and fusion models. The results also highlight the effectiveness of the predictive transition mechanism between the normal and planned events and also the event information representation.
Rights
Permission to publish the abstract has been given by SAGE, copyright remains with them.
Recommended Citation
Zhao, Y., & Ma, Z. (2022). Naïve Bayes-Based Transition Model for Short-Term Metro Passenger Flow Prediction under Planned Events. Transportation Research Record: Journal of the Transportation Research Board, Vol. 2676(9), pp. 309-324.
