Multi-output Deep Learning for Bus Arrival Time Predictions

Research output: Contribution to journalConference articleResearchpeer-review

11 Downloads (Pure)

Abstract

Accurate and reliable predictions for bus arrival in public transport networks are essential for delivering an attractive service. This paper presents a multi-model approach for bus arrival prediction. The approach uses three distinct sub-models in an ensemble model. A multi-output, multi-time-step, deep neural network using Convolutional and Long short-term memory (LSTM) layers is used for travel time, and more simplistic models are used for dwell time and seasonal components. The method is empirically evaluated and compared to other popular approaches. We find that the proposed model saturations outperforms the other methods, while in other saturations performs similar.
Original languageEnglish
JournalTransportation Research Procedia
Volume41
Pages (from-to)138-145
ISSN2352-1465
DOIs
Publication statusPublished - 2019
EventInternational Scientific Conference on Mobility and Transport Urban Mobility - Munich, Germany
Duration: 13 Jun 201814 Jun 2018

Conference

ConferenceInternational Scientific Conference on Mobility and Transport Urban Mobility
CountryGermany
CityMunich
Period13/06/201814/06/2018

Cite this

@inproceedings{4c6767610ad448489c67bcdcec9df81d,
title = "Multi-output Deep Learning for Bus Arrival Time Predictions",
abstract = "Accurate and reliable predictions for bus arrival in public transport networks are essential for delivering an attractive service. This paper presents a multi-model approach for bus arrival prediction. The approach uses three distinct sub-models in an ensemble model. A multi-output, multi-time-step, deep neural network using Convolutional and Long short-term memory (LSTM) layers is used for travel time, and more simplistic models are used for dwell time and seasonal components. The method is empirically evaluated and compared to other popular approaches. We find that the proposed model saturations outperforms the other methods, while in other saturations performs similar.",
author = "Petersen, {Niklas Christoffer} and Filipe Rodrigues and Pereira, {Francisco Camara}",
year = "2019",
doi = "10.1016/j.trpro.2019.09.025",
language = "English",
volume = "41",
pages = "138--145",
journal = "Transportation Research Procedia",
issn = "2352-1465",
publisher = "Elsevier",

}

Multi-output Deep Learning for Bus Arrival Time Predictions. / Petersen, Niklas Christoffer; Rodrigues, Filipe; Pereira, Francisco Camara.

In: Transportation Research Procedia, Vol. 41, 2019, p. 138-145.

Research output: Contribution to journalConference articleResearchpeer-review

TY - GEN

T1 - Multi-output Deep Learning for Bus Arrival Time Predictions

AU - Petersen, Niklas Christoffer

AU - Rodrigues, Filipe

AU - Pereira, Francisco Camara

PY - 2019

Y1 - 2019

N2 - Accurate and reliable predictions for bus arrival in public transport networks are essential for delivering an attractive service. This paper presents a multi-model approach for bus arrival prediction. The approach uses three distinct sub-models in an ensemble model. A multi-output, multi-time-step, deep neural network using Convolutional and Long short-term memory (LSTM) layers is used for travel time, and more simplistic models are used for dwell time and seasonal components. The method is empirically evaluated and compared to other popular approaches. We find that the proposed model saturations outperforms the other methods, while in other saturations performs similar.

AB - Accurate and reliable predictions for bus arrival in public transport networks are essential for delivering an attractive service. This paper presents a multi-model approach for bus arrival prediction. The approach uses three distinct sub-models in an ensemble model. A multi-output, multi-time-step, deep neural network using Convolutional and Long short-term memory (LSTM) layers is used for travel time, and more simplistic models are used for dwell time and seasonal components. The method is empirically evaluated and compared to other popular approaches. We find that the proposed model saturations outperforms the other methods, while in other saturations performs similar.

U2 - 10.1016/j.trpro.2019.09.025

DO - 10.1016/j.trpro.2019.09.025

M3 - Conference article

VL - 41

SP - 138

EP - 145

JO - Transportation Research Procedia

JF - Transportation Research Procedia

SN - 2352-1465

ER -