Crash Prediction using Ensemble Methods

Mohammed Ameksa, Hajar Mousannif, Hassan Al Moatassime, Zouhair Elamrani Abou Elassad

2021

Abstract

The prediction of traffic accidents is a major concern worldwide due to its negative impact on all sectors. The human and financial losses caused by road accidents have become increasingly important. This study aims to investigate the prediction of an accident using several ensemble-based methods, including the GBM gradient boosting machine, the XGB extreme gradient boosting, and RF random forest. To achieve this, we used driver’ inputs data extracted from the outputs of a driving simulator located at Cadi Ayyad University UCA. And the evaluation of the developed models was carried out for both configurations, before and after tuning of hyperparameters. Results show that the RF outperforms GBM and XGB for the default parameters with an accuracy of 93%. However, after hyperparameters optimization, it had been noticed that even if the algorithms are not the same, their performances are almost equal. The highest performance was performed after tuning of hyperparameters.

Paper Citation

in Harvard Style

Ameksa M., Mousannif H., Al Moatassime H. and Elamrani Abou Elassad Z. (2021). Crash Prediction using Ensemble Methods. In Proceedings of the 2nd International Conference on Big Data, Modelling and Machine Learning - Volume 1: BML, ISBN 978-989-758-559-3, pages 211-215. DOI: 10.5220/0010731200003101

in Bibtex Style

@conference{bml21,
author={Mohammed Ameksa and Hajar Mousannif and Hassan Al Moatassime and Zouhair Elamrani Abou Elassad},
title={Crash Prediction using Ensemble Methods},
booktitle={Proceedings of the 2nd International Conference on Big Data, Modelling and Machine Learning - Volume 1: BML,},
year={2021},
pages={211-215},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0010731200003101},
isbn={978-989-758-559-3},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 2nd International Conference on Big Data, Modelling and Machine Learning - Volume 1: BML,
TI - Crash Prediction using Ensemble Methods
SN - 978-989-758-559-3
AU - Ameksa M.
AU - Mousannif H.
AU - Al Moatassime H.
AU - Elamrani Abou Elassad Z.
PY - 2021
SP - 211
EP - 215
DO - 10.5220/0010731200003101