Table 1. The summary of previous literatures related to accident severity.
Author | Goal | Influencing factors considered | Severity considered | Models | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
A | B | C | I | II | III | IV | V | VI | VII | VIII | IX | |||
Sattar et al. [10] | √ | √ | √ | √ | √ | √ | √ | severe, non-severe | GNN, RF, XGBoost, ANN | |||||
Yan et al. [8] | √ | √ | √ | √ | √ | √ | √ | √ | non-fatal, fatal | LightGBM | ||||
Mohammadpour et al. [28] | √ | √ | √ | √ | √ | √ | √ | √ | fatal and severe injury, less severe injury, PDO | RF, KNN, GBDT, SVM, Multi-Layer Perceptron | ||||
Hyodo et al. [11] | √ | √ | √ | √ | √ | √ | √ | √ | minor, severe, or fatal | Ordered Probit Model | ||||
Hosseinzadeh et al. [6] | √ | √ | √ | √ | √ | √ | √ | √ | fatal, non-fatal | SVM, random parameter logit model | ||||
Li et al. [5] | √ | √ | √ | √ | √ | √ | √ | √ | √ | incapacitating crash, fatal crash | RF, GBDT, AdaBoost, Mixed Logit | |||
Islam et al. [7] | √ | √ | √ | √ | √ | √ | √ | fatal, non-fatal crashes | multinomial logit model | |||||
Ahmed et al. [18] | √ | √ | √ | √ | √ | √ | √ | √ | √ | fatal, serious, minor, and non-injury | RF, Decision Jungle, AdaBoost, XGBoost, LightGBM, CatBoost | |||
Shiran et al. [13] | √ | √ | √ | √ | √ | √ | √ | √ | PDO, fatality, severe injury, other visible injuries, and complaint of pain | MNL, ANN-MLP, CHAID, and C5.0 | ||||
Zainuddin et al. [16] | √ | √ | √ | √ | √ | √ | non-fatal or fatal accident. | Descriptive and chi-square test | ||||||
Ahmed et al. [18] | √ | √ | √ | √ | √ | √ | √ | √ | √ | fatal, serious, minor, and non-injury crashes | RF, Decision Jungle, AdaBoost, XGBoost, LightGBM, CatBoost | |||
Zhou et al. [29] | √ | √ | √ | √ | √ | √ | √ | √ | no injury, injury, fatality | MNL, Naive Bayes, SVM, and XGBoost | ||||
Panda et al. [30] | √ | √ | √ | √ | √ | √ | √ | killed and injured | SVM, RF, GBDT, XGBoost | |||||
Yang et al. [22] | √ | √ | √ | √ | √ | property loss, Injuries, Fatal | XGBoost + Bayesian network model | |||||||
Mahashhash et al. [24] | √ | √ | √ | √ | √ | √ | non-severe injury, Severe injury or fatal | Binary logit model | ||||||
Lee et al. [31] | √ | √ | √ | √ | √ | √ | √ | √ | PDO, bodily damage | Logit Model |
A. Identify the key contributing factors; B. Examine the relationships between factors and accident severity; C. Compare several model’s performance.
I. Driver-related and driving behavior factors. II. Vehicle-related factors. III. Road geometry factors. IV. Traffic characteristic factors V. Weather factors. VI. Lighting conditions. VII. Crash characteristic factors. VIII. Temporal variables. IX. Other environment factors, such as built environment, spatial configuration.