Subway Passenger Flow Prediction for Special Events Using Smart Card Data
- 29 May 2019
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Transportation Systems
- Vol. 21 (3), 1109-1120
- https://doi.org/10.1109/tits.2019.2902405
Abstract
In order to reduce passenger delays and prevent severe overcrowding in the subway system, it is necessary to accurately predict the short-term passenger flow during special events. However, few studies have been conducted to predict the subway passenger flow under these conditions. Traditional methods, such as the autoregressive integrated moving average (ARIMA) model, were commonly used to analyze regular traffic demands. These methods usually neglected the volatility (heteroscedasticity) in passenger flow influenced by unexpected external factors. This paper, therefore, proposed a generic framework to analyze short-term passenger flow, considering the dynamic volatility and nonlinearity of passenger flow during special events. Four different generalized autoregressive conditional heteroscedasticity models, along with the ARIMA model, were used to model the mean and volatility of passenger flow based on the transit smart card data from two stations near the Olympic Sports Center, Nanjing, China. Multiple statistical methods were applied to evaluate the performance of the hybrid models. The results indicate that the volatility of passenger flow had significant nonlinear and asymmetric features during special events. The proposed framework could effectively capture the mean and volatility of passenger flow, and outperform the traditional methods in terms of accuracy and reliability. Overall, this paper can help transit agencies to better understand the deterministic and stochastic changes of the passenger flow, and implement precautionary countermeasures for large crowds in a subway station before and after special events.Keywords
Funding Information
- National High-tech Research and Development Program (2014AA110303)
This publication has 39 references indexed in Scilit:
- Temporal aggregation in traffic data: implications for statistical characteristics and model choiceTransportation Letters, 2011
- Short-Term Traffic Flow Forecasting Using Fuzzy Logic System MethodsJournal of Intelligent Transportation Systems, 2008
- Illuminating the “black box”: a randomization approach for understanding variable contributions in artificial neural networksEcological Modelling, 2002
- Short-Term Traffic Flow Prediction Using Neuro-Genetic AlgorithmsJournal of Intelligent Transportation Systems, 2002
- On the Relation between the Expected Value and the Volatility of the Nominal Excess Return on StocksThe Journal of Finance, 1993
- Conditional Heteroskedasticity in Asset Returns: A New ApproachEconometrica, 1991
- Generalized autoregressive conditional heteroskedasticityJournal of Econometrics, 1986
- Autoregressive Conditional Heteroscedasticity with Estimates of the Variance of United Kingdom InflationEconometrica, 1982
- A Three-Stage Iterative Procedure for Space-Time ModelingTechnometrics, 1980
- Some Comments on a Paper by Chatfield and Prothero and on A Review by KendallJournal of the Royal Statistical Society. Series A (General), 1973