Open AccessPrediction of Metro Short-term Entry Flow Based on Passenger Flow Characteristic Analysis
Author(s) -
Shengchuan Zhao,
Chongyu Zhang,
Huasheng Liu
Publication year - 2020
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.179
H-Index - 26
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/608/1/012021
Subject(s) - mean absolute percentage error , term (time) , mean squared error , artificial neural network , random forest , support vector machine , computer science , flow (mathematics) , stability (learning theory) , regression analysis , statistics , engineering , mathematics , artificial intelligence , machine learning , physics , geometry , quantum mechanics
Prediction of short-term incoming passenger flow at metro stations is of great significance to the stable operation of Metro networks. Taking AFC data of Chengdu Metro as data source, based on Stochastic Forest model, support vector machine regression model and neural network model in machine learning method, this paper makes short-term prediction of metro entry flow and comparative analysis of three models. Metro stations are divided into four types according to the passenger flow characteristics: residential type, commercial type, business type and terminal type. Three models are trained in workdays and weekends conditions for different types of stations. The average absolute percentage error (MAPE) and root mean square error (RMSE) were used to evaluate the accuracy and stability of the prediction results. The results show that BP neural network has the best comprehensive performance and random forest has better prediction accuracy for stations with strong periodicity.