Passenger flow prediction in urban rail transit is a fundamental task in rail transit construction, and accurate passenger flow prediction can avoid the risk of waste or shortage of rail operation resources. The article proposes to apply XGBoost, an integration-based machine learning method, to rail passenger flow prediction. Taking Xi’an Metro Line 2 as the research object, the swipe card data from April to May 2017 for the whole line was used. The passenger flow characteristics of special holidays, weekdays and rest days were analysed, and the study time period was determined to be a weekday with more concentrated passenger flow distribution. The prediction results of XGBoost were compared and analysed with those of BP neural network model and ARMA model, and it was found that the XGBoost algorithm has higher prediction accuracy and can provide reference for rail transit construction prediction.
Urban Rail Transit Passenger Flow Forecasting—XGBoost
22nd COTA International Conference of Transportation Professionals ; 2022 ; Changsha, Hunan Province, China
CICTP 2022 ; 1142-1150
2022-09-08
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Multi-graph convolutional network for short-term passenger flow forecasting in urban rail transit
IET | 2020
|Multi‐graph convolutional network for short‐term passenger flow forecasting in urban rail transit
Wiley | 2020
|Urban rail transit passenger flow prediction method under emergency
Europäisches Patentamt | 2023
|