Citation

Lee J, Yoon T, Kwon S, Lee J. Appl. Sci. (Basel) 2020; 10(1): e129.

Copyright

(Copyright © 2020, MDPI: Multidisciplinary Digital Publishing Institute)

DOI

10.3390/app10010129

PMID

unavailable

Abstract

There have been numerous studies on traffic accidents and their severity, particularly in relation to weather conditions and road geometry. In these studies, traditional statistical methods have been employed, such as linear regression, logistic regression, and negative binomial regression modeling, which are the most common linear and non-linear regression analysis methods. In this research, machine learning architecture was applied to this problem using the random forest, artificial neural network, and decision tree techniques to ascertain the strengths and weaknesses of these methods. Three data sets were used: road geometry data, precipitation data, and traffic accident data over nine years corresponding to the Naebu Expressway, which is located in Seoul, Korea. For the model evaluation, three measures were employed: the out-of-bag estimate of error rate (OOB), mean square error (MSE), and root mean square error (RMSE). The low mean OOB, MSE, and RMSE observed in the results obtained using the proposed random forest model demonstrate its accuracy.

Language: en

Keywords

accident severity level; artificial neural network; decision tree algorithm; machine learning architecture; random forest model; road hazard zone forecasting; road surface condition