A Comparison of Approaches for the Time Series Forecasting of Motorway Traffic Flow Rate at Hourly and Daily Aggregation Levels

Bas Van Der Bijl; Bart Gijsbertsen; Stan Van Loon; Yorran Reurich; Tom De Valk; Thomas Koch; Elenna Dugundji

doi:10.1016/j.procs.2022.03.030

A Comparison of Approaches for the Time Series Forecasting of Motorway Traffic Flow Rate at Hourly and Daily Aggregation Levels

Bas Van Der Bijl, Bart Gijsbertsen, Stan Van Loon, Yorran Reurich, Tom De Valk, Thomas Koch^*, Elenna Dugundji

^*Corresponding author for this work

Mathematics

Research output: Contribution to Journal › Article › Academic › peer-review

Abstract

Congestion forms a large problem in many major metropolitan regions around the world, leading to delays and societal costs. Congestion is generally associated with reduced average speed at a high traffic flow rate. This traffic flow rate is defined as the number of vehicles that pass a certain location at a given time. The modelling and prediction of this traffic flow rate may lead to valuable insights that may be used to reduce congestion and societal costs. This study aims to predict the traffic flow rate for 41 different locations in and around Amsterdam, The Netherlands. Using TBATS, SARIMAX and LSTM models, among others, the traffic flow rate of these locations has successfully been modelled. These models may provide accurate predictions for the future flow rate, which may be useful for the identification of infrastructure bottlenecks and the scheduling of maintenance. Considering the dramatic effects of the COVID-19 pandemic on the traffic flow rate, the inclusion of 2020 data with a number of external factors, could lead to an improvement of the results and the ability to model the future effects of the pandemic.

Original language	English
Pages (from-to)	213-222
Number of pages	10
Journal	Procedia Computer Science
Volume	201
Issue number	C
Early online date	27 Apr 2022
DOIs	https://doi.org/10.1016/j.procs.2022.03.030
Publication status	Published - 2022
Event	13th International Conference on Ambient Systems, Networks and Technologies, ANT 2022 / 5th International Conference on Emerging Data and Industry 4.0, EDI40 2022 - Porto, Portugal Duration: 22 Mar 2022 → 25 Mar 2022

Bibliographical note

Part of special issue: The 13th International Conference on Ambient Systems, Networks and Technologies (ANT) / The 5th International Conference on Emerging Data and Industry 4.0 (EDI40), Edited by Elhadi Shakshuki.

Funding Information:
This research has been conducted in the framework of the Impact Study North/Southline research theme ”Mobility and Accessiblity” hosted at CWI and funded in part by the Municipality of Amsterdam and the regional transportation authority of Amsterdam. We would like to thank the Faculty of Science at the Vrije Universiteit Amsterdam for giving us the opportunity to initially work on the research during the Project Optimization of Business Processes (POBP).

Publisher Copyright:
© 2022 Elsevier B.V.. All rights reserved.

Keywords

machine learning
prediction
time series
traffic flow
traffic management

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1016/j.procs.2022.03.030

Persistent URL (handle)

Persistent link

Cite this

@article{4f7e099626b545e4a4b05fb3b3d02051,

title = "A Comparison of Approaches for the Time Series Forecasting of Motorway Traffic Flow Rate at Hourly and Daily Aggregation Levels",

abstract = "Congestion forms a large problem in many major metropolitan regions around the world, leading to delays and societal costs. Congestion is generally associated with reduced average speed at a high traffic flow rate. This traffic flow rate is defined as the number of vehicles that pass a certain location at a given time. The modelling and prediction of this traffic flow rate may lead to valuable insights that may be used to reduce congestion and societal costs. This study aims to predict the traffic flow rate for 41 different locations in and around Amsterdam, The Netherlands. Using TBATS, SARIMAX and LSTM models, among others, the traffic flow rate of these locations has successfully been modelled. These models may provide accurate predictions for the future flow rate, which may be useful for the identification of infrastructure bottlenecks and the scheduling of maintenance. Considering the dramatic effects of the COVID-19 pandemic on the traffic flow rate, the inclusion of 2020 data with a number of external factors, could lead to an improvement of the results and the ability to model the future effects of the pandemic.",

keywords = "machine learning, prediction, time series, traffic flow, traffic management",

author = "{Van Der Bijl}, Bas and Bart Gijsbertsen and {Van Loon}, Stan and Yorran Reurich and {De Valk}, Tom and Thomas Koch and Elenna Dugundji",

note = "Part of special issue: The 13th International Conference on Ambient Systems, Networks and Technologies (ANT) / The 5th International Conference on Emerging Data and Industry 4.0 (EDI40), Edited by Elhadi Shakshuki. Funding Information: This research has been conducted in the framework of the Impact Study North/Southline research theme ”Mobility and Accessiblity” hosted at CWI and funded in part by the Municipality of Amsterdam and the regional transportation authority of Amsterdam. We would like to thank the Faculty of Science at the Vrije Universiteit Amsterdam for giving us the opportunity to initially work on the research during the Project Optimization of Business Processes (POBP). Publisher Copyright: {\textcopyright} 2022 Elsevier B.V.. All rights reserved.; 13th International Conference on Ambient Systems, Networks and Technologies, ANT 2022 / 5th International Conference on Emerging Data and Industry 4.0, EDI40 2022 ; Conference date: 22-03-2022 Through 25-03-2022",

year = "2022",

doi = "10.1016/j.procs.2022.03.030",

language = "English",

volume = "201",

pages = "213--222",

journal = "Procedia Computer Science",

issn = "1877-0509",

publisher = "Elsevier BV",

number = "C",

}

TY - JOUR

T1 - A Comparison of Approaches for the Time Series Forecasting of Motorway Traffic Flow Rate at Hourly and Daily Aggregation Levels

AU - Van Der Bijl, Bas

AU - Gijsbertsen, Bart

AU - Van Loon, Stan

AU - Reurich, Yorran

AU - De Valk, Tom

AU - Koch, Thomas

AU - Dugundji, Elenna

N1 - Part of special issue: The 13th International Conference on Ambient Systems, Networks and Technologies (ANT) / The 5th International Conference on Emerging Data and Industry 4.0 (EDI40), Edited by Elhadi Shakshuki. Funding Information: This research has been conducted in the framework of the Impact Study North/Southline research theme ”Mobility and Accessiblity” hosted at CWI and funded in part by the Municipality of Amsterdam and the regional transportation authority of Amsterdam. We would like to thank the Faculty of Science at the Vrije Universiteit Amsterdam for giving us the opportunity to initially work on the research during the Project Optimization of Business Processes (POBP). Publisher Copyright: © 2022 Elsevier B.V.. All rights reserved.

PY - 2022

Y1 - 2022

N2 - Congestion forms a large problem in many major metropolitan regions around the world, leading to delays and societal costs. Congestion is generally associated with reduced average speed at a high traffic flow rate. This traffic flow rate is defined as the number of vehicles that pass a certain location at a given time. The modelling and prediction of this traffic flow rate may lead to valuable insights that may be used to reduce congestion and societal costs. This study aims to predict the traffic flow rate for 41 different locations in and around Amsterdam, The Netherlands. Using TBATS, SARIMAX and LSTM models, among others, the traffic flow rate of these locations has successfully been modelled. These models may provide accurate predictions for the future flow rate, which may be useful for the identification of infrastructure bottlenecks and the scheduling of maintenance. Considering the dramatic effects of the COVID-19 pandemic on the traffic flow rate, the inclusion of 2020 data with a number of external factors, could lead to an improvement of the results and the ability to model the future effects of the pandemic.

AB - Congestion forms a large problem in many major metropolitan regions around the world, leading to delays and societal costs. Congestion is generally associated with reduced average speed at a high traffic flow rate. This traffic flow rate is defined as the number of vehicles that pass a certain location at a given time. The modelling and prediction of this traffic flow rate may lead to valuable insights that may be used to reduce congestion and societal costs. This study aims to predict the traffic flow rate for 41 different locations in and around Amsterdam, The Netherlands. Using TBATS, SARIMAX and LSTM models, among others, the traffic flow rate of these locations has successfully been modelled. These models may provide accurate predictions for the future flow rate, which may be useful for the identification of infrastructure bottlenecks and the scheduling of maintenance. Considering the dramatic effects of the COVID-19 pandemic on the traffic flow rate, the inclusion of 2020 data with a number of external factors, could lead to an improvement of the results and the ability to model the future effects of the pandemic.

KW - machine learning

KW - prediction

KW - time series

KW - traffic flow

KW - traffic management

UR - http://www.scopus.com/inward/record.url?scp=85132140153&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85132140153&partnerID=8YFLogxK

U2 - 10.1016/j.procs.2022.03.030

DO - 10.1016/j.procs.2022.03.030

M3 - Article

AN - SCOPUS:85132140153

SN - 1877-0509

VL - 201

SP - 213

EP - 222

JO - Procedia Computer Science

JF - Procedia Computer Science

IS - C

T2 - 13th International Conference on Ambient Systems, Networks and Technologies, ANT 2022 / 5th International Conference on Emerging Data and Industry 4.0, EDI40 2022

Y2 - 22 March 2022 through 25 March 2022

ER -

A Comparison of Approaches for the Time Series Forecasting of Motorway Traffic Flow Rate at Hourly and Daily Aggregation Levels

Abstract

Bibliographical note

Keywords

UN SDGs

Access to Document

Persistent URL (handle)

Other files and links

Fingerprint

Cite this