Citation

Li MK, Yu JJ, Ma L, Zhang W. Accid. Anal. Prev. 2019; 123: 79-87.

Affiliation

Department of Industrial Engineering, Tsinghua University, Beijing, 100084, China; State Key Laboratory of Automotive Safety and Energy, Tsinghua University, Beijing, 100084, China. Electronic address: zhangwei@tsinghua.edu.cn.

Copyright

(Copyright © 2019, Elsevier Publishing)

DOI

10.1016/j.aap.2018.11.001

PMID

30468949

Abstract

Taxi drivers worldwide often have very long driving hours and experience frequent fatigue. These conditions are associated with a high prevalence of fatigue and accidents. However, the key factors that distinguish high/low fatigue-related accident risk (FRAR) taxi drivers are uncertain. By examining a series of potential factors related with fatigue or accident risk as discussed in previous research, the objective was to find out the most important factors that relate to taxi driver's FRAR, and to investigate the association of these factors and taxi driver's FRAR. Modeling methods were applied to questionnaire data collected from Beijing taxi drivers. A 269-sample dataset was analyzed to identify key factors related to FRAR and to fit FRAR prediction models. The model's performance on high-risk driver prediction was then tested using another independently collected 100-sample dataset. High-risk taxi drivers had significantly longer driving hours per working day, lower rest ratios, less driving experience, and were more confident about their fatigue resistance. The FRAR model with only four major measurable predictors achieved a sensitivity of 91.9% and a specificity of 94.6% on predicting labeled data. Adjusting drive-rest habits and self-evaluation pertaining to these predictors is good for high-risk drivers to mitigate their accident risk. It was concluded that taxi drivers' drive-rest habits, experience, and intention for fatigue driving are crucial, and to a large degree determine their FRAR, and the prediction model can satisfactorily identify high-risk taxi drivers.

Copyright © 2018 Elsevier Ltd. All rights reserved.

Language: en

Keywords

Accident analysis; Driver behavior; Fatigue; Modeling; Risk prediction