Open AccessShort‐term passenger flow forecast of urban rail transit based on GPR and KRR
Author(s) -
Guo Zhiqiang,
Zhao Xin,
Chen Yaxin,
Wu Wei,
Yang Jie
Publication year - 2019
Publication title -
iet intelligent transport systems
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.579
H-Index - 45
eISSN - 1751-9578
pISSN - 1751-956X
DOI - 10.1049/iet-its.2018.5530
Subject(s) - computer science , term (time) , ground penetrating radar , urban rail transit , feature selection , feature (linguistics) , process (computing) , data mining , stability (learning theory) , engineering , artificial intelligence , machine learning , radar , transport engineering , telecommunications , physics , quantum mechanics , linguistics , philosophy , operating system
Short‐term passenger flow forecasting can help the operation management department to adjust the related work. At the same time, it can also guide the traveller to choose a reasonable travel time and route, which plays an important role in promoting the development and construction of the city. In this study, the authors propose a hybrid prediction model based on kernel ridge regression (KRR) and Gaussian process regression (GPR) to predict the short‐term passenger flow of urban rail transit, and verify it on the Automatic Fare Collection System (AFC) dataset. Firstly, they utilise the stability feature selection algorithm to control the error of finite samples and use a GPR algorithm to obtain the original result. Then, they introduce stacked auto‐encoder network to construct a feature extraction model, and apply k ‐means method to divide the stations into different types, defining as a site feature. Furthermore, they choose KRR algorithm with the combination of GPR prediction result and the holiday information, the station category information mentioned above, achieving the final prediction. The algorithm proposed in this study effectively improves the prediction accuracy and ensures time efficiency, and all the indicators are better than the existing algorithms.