Although smart-card data secures collective travel information on public transportation users, the reality is that only a few cities are equipped with an automatic fare collection (AFC) system that can provide user information for both boarding and alighting locations. Many researchers have delved into forecasting the destinations of smart-card users. Such effort, however, have never been validated with actual data on a large scale. In the present study, a deep-learning model was developed to estimate the destinations of bus passengers based on both entry-only smart-card data and land-use characteristics. A supervised machine-learning model was trained using exact information on both boarding and alighting. That information was provided by the AFC system in Seoul, Korea. The model performance was superior to that of the most prevalent schemes developed thus far.
Deep-learning architecture to forecast destinations of bus passengers from entry-only smart-card data
IET Intelligent Transport Systems ; 11 , 6 ; 334-339
2017-08-01
6 pages
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
user information , traffic information systems , deep-learning architecture , smart cards , boarding locations , alighting locations , learning (artificial intelligence) , destinations forecasting , public transportation users , bus passengers , collective travel information , public transport , supervised machine-learning model , AFC system , entry-only smart-card data , land-use characteristics , automatic fare collection system
Metadata by IET is licensed under CC BY 3.0
Wiley | 2017
|Deduction of Passengers' Route Choices From Smart Card Data
Online Contents | 2015
|A Deep Learning Approach to Infer Employment Status of Passengers by Using Smart Card Data
IEEE | 2020
|