Crash Injury Severity Prediction Using an Ordinal Classification Machine Learning Approach

In many related works, nominal classification algorithms ignore the order between injury severity levels and make sub-optimal predictions. Existing ordinal classification methods suffer rank inconsistency and rank non-monotonicity. The aim of this paper is to propose an ordinal classification approa...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Shengxue Zhu, Ke Wang, Chongyi Li
Formato:	article
Lenguaje:	EN
Publicado:	MDPI AG 2021
Materias:	crash severity ordinal classification imbalance data machine learning sampling Medicine R
Acceso en línea:	https://doaj.org/article/3b6e2cb0b88b4ac1a903d470968771d1
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:3b6e2cb0b88b4ac1a903d470968771d1
record_format	dspace
spelling	oai:doaj.org-article:3b6e2cb0b88b4ac1a903d470968771d12021-11-11T16:40:35ZCrash Injury Severity Prediction Using an Ordinal Classification Machine Learning Approach10.3390/ijerph1821115641660-46011661-7827https://doaj.org/article/3b6e2cb0b88b4ac1a903d470968771d12021-11-01T00:00:00Zhttps://www.mdpi.com/1660-4601/18/21/11564https://doaj.org/toc/1661-7827https://doaj.org/toc/1660-4601In many related works, nominal classification algorithms ignore the order between injury severity levels and make sub-optimal predictions. Existing ordinal classification methods suffer rank inconsistency and rank non-monotonicity. The aim of this paper is to propose an ordinal classification approach to predict traffic crash injury severity and to test its performance over existing machine learning classification methods. First, we compare the performance of the neural network, XGBoost, and SVM classifiers in injury severity prediction. Second, we utilize a severity category-combination method with oversampling to relieve the class-imbalance problem prevalent in crash data. Third, we take advantage of probability calibration and the optimal probability threshold moving to improve the prediction ability of ordinal classification. The proposed approach can satisfy the rank consistency and rank monotonicity requirement and is proved to be superior to other ordinal classification methods and nominal classification machine learning by statistical significance test. Important factors relating to injury severity are selected based on their permutation feature importance scores. We find that converting severity levels into three classes, minor injury, moderate injury, and serious injury, can substantially improve the prediction precision.Shengxue ZhuKe WangChongyi LiMDPI AGarticlecrash severityordinal classificationimbalance datamachine learningsamplingMedicineRENInternational Journal of Environmental Research and Public Health, Vol 18, Iss 11564, p 11564 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	crash severity ordinal classification imbalance data machine learning sampling Medicine R
spellingShingle	crash severity ordinal classification imbalance data machine learning sampling Medicine R Shengxue Zhu Ke Wang Chongyi Li Crash Injury Severity Prediction Using an Ordinal Classification Machine Learning Approach
description	In many related works, nominal classification algorithms ignore the order between injury severity levels and make sub-optimal predictions. Existing ordinal classification methods suffer rank inconsistency and rank non-monotonicity. The aim of this paper is to propose an ordinal classification approach to predict traffic crash injury severity and to test its performance over existing machine learning classification methods. First, we compare the performance of the neural network, XGBoost, and SVM classifiers in injury severity prediction. Second, we utilize a severity category-combination method with oversampling to relieve the class-imbalance problem prevalent in crash data. Third, we take advantage of probability calibration and the optimal probability threshold moving to improve the prediction ability of ordinal classification. The proposed approach can satisfy the rank consistency and rank monotonicity requirement and is proved to be superior to other ordinal classification methods and nominal classification machine learning by statistical significance test. Important factors relating to injury severity are selected based on their permutation feature importance scores. We find that converting severity levels into three classes, minor injury, moderate injury, and serious injury, can substantially improve the prediction precision.
format	article
author	Shengxue Zhu Ke Wang Chongyi Li
author_facet	Shengxue Zhu Ke Wang Chongyi Li
author_sort	Shengxue Zhu
title	Crash Injury Severity Prediction Using an Ordinal Classification Machine Learning Approach
title_short	Crash Injury Severity Prediction Using an Ordinal Classification Machine Learning Approach
title_full	Crash Injury Severity Prediction Using an Ordinal Classification Machine Learning Approach
title_fullStr	Crash Injury Severity Prediction Using an Ordinal Classification Machine Learning Approach
title_full_unstemmed	Crash Injury Severity Prediction Using an Ordinal Classification Machine Learning Approach
title_sort	crash injury severity prediction using an ordinal classification machine learning approach
publisher	MDPI AG
publishDate	2021
url	https://doaj.org/article/3b6e2cb0b88b4ac1a903d470968771d1
work_keys_str_mv	AT shengxuezhu crashinjuryseveritypredictionusinganordinalclassificationmachinelearningapproach AT kewang crashinjuryseveritypredictionusinganordinalclassificationmachinelearningapproach AT chongyili crashinjuryseveritypredictionusinganordinalclassificationmachinelearningapproach
_version_	1718432232961474560

Crash Injury Severity Prediction Using an Ordinal Classification Machine Learning Approach

Ejemplares similares