Traffic Flow Prediction from Loop Counter Sensor Data using Machine Learning Methods

Blaž Kažic, Dunja Mladenić, Aljaž Košmerlj

2015

Abstract

Due to increasing demand and growing cities, traffic prediction has been a topic of interest for many researchers for the past few decades. The availability of large amounts of traffic-related data and the emerging field of machine learning algorithms has led to a significant leap towards data-driven methods. In this paper, loop counter data are used to develop models that can predict traffic flow for several different prediction intervals into the future. In depth exploratory data analysis and statistical testing is performed to obtain good quality informative features. Several feature sets were compared by using different machine learning methods: Ridge Regression, SVR and Random Forests. The results show that in order to obtain good prediction results thorough feature extraction is just as or even more important than learning method selection. Good features enables us to use less complex methods, which run faster, are more reliable and easier to maintain. In conclusion, we address ideas regarding how predictions could be improved even further.

Paper Citation

in Harvard Style

Kažic B., Mladenić D. and Košmerlj A. (2015). Traffic Flow Prediction from Loop Counter Sensor Data using Machine Learning Methods . In Proceedings of the 1st International Conference on Vehicle Technology and Intelligent Transport Systems - Volume 1: VEHITS, ISBN 978-989-758-109-0, pages 119-127. DOI: 10.5220/0005495001190127

in Bibtex Style

@conference{vehits15,
author={Blaž Kažic and Dunja Mladenić and Aljaž Košmerlj},
title={Traffic Flow Prediction from Loop Counter Sensor Data using Machine Learning Methods},
booktitle={Proceedings of the 1st International Conference on Vehicle Technology and Intelligent Transport Systems - Volume 1: VEHITS,},
year={2015},
pages={119-127},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0005495001190127},
isbn={978-989-758-109-0},
}

in EndNote Style

TY - CONF
JO - Proceedings of the 1st International Conference on Vehicle Technology and Intelligent Transport Systems - Volume 1: VEHITS,
TI - Traffic Flow Prediction from Loop Counter Sensor Data using Machine Learning Methods
SN - 978-989-758-109-0
AU - Kažic B.
AU - Mladenić D.
AU - Košmerlj A.
PY - 2015
SP - 119
EP - 127
DO - 10.5220/0005495001190127