Individual trip prediction on multi-mode transit system by identifying transit mode and trip regularity using smart card data

Publication Type

Journal Article

Publication Date

1-2026

Abstract

With the rapid increase in population and busy urban lifestyle, individual user's trip prediction by Bus Rapid Transit (BRT) and Mass Rapid Transit (MRT) in a city has several applications. Nowadays, users use smart card data to access the BRT and MRT systems and the dataset generated from the use of smart cards by users can be utilized for individual user's trip prediction. Most of the existing methods focus on the prediction of passenger flow or destination. However, in this study, we focus on individual user's trip predictions. Specifically, we focus on two prediction problems: (1) when the user starts a current trip, we want to predict the destination of the current trip, and (2) when and where the user will start the next trip, i.e. the time interval between end of last trip and start of next trip, and origin of the next trip. To tackle these problems, we propose a framework that provides the methods for these two kinds of prediction related to an individual user's trip: (1) prediction of destination of the current trip by the proposed regular–irregular module, and (2) prediction of duration between the last trip and beginning of next trip, i.e. when the next trip will start and the origin of the next trip, i.e. where the next trip will start. We use smart card data and several features related to the trip information. Extensive experiments on a real dataset (Singapore EZ-Link smart card dataset) demonstrate that the proposed method outperforms the competitors.

Keywords

Individual mobility, smart urban planning, next-trip prediction, smart card data, multi-system transportation

Discipline

Artificial Intelligence and Robotics | Operations Research, Systems Engineering and Industrial Engineering | Transportation

Research Areas

Intelligent Systems and Optimization

Publication

International Journal of Urban Sciences

First Page

1

Last Page

18

ISSN

1226-5934

Identifier

10.1080/12265934.2026.2614035

Publisher

Taylor & Francis

Additional URL

https://doi.org/10.1080/12265934.2026.2614035

This document is currently not available here.

Share

COinS