Fig 1.
Operational framework of proposed Boruta algorithm and machine learning classification system for injury severity.
Table 1.
Number of injuries in single and multiple-vehicle accidents.
Fig 2.
Wrapper-based Boruta Algorithm (BA).
Fig 3.
Confusion matrix.
Fig 4.
(a) Box plots of attributes based on importance values. (b) Importance value of attributes in each classifier run.
Table 2.
Attributes importance statistics by Boruta Algorithm (BA).
Fig 5.
Confusion matrix of selected important attributes and all attributes.
Table 3.
Comparison of prediction performance of different classifiers for single and multiple vehicles accidents.
Fig 6.
Confusion matrix for single-vehicle accidents using various classifiers.
Fig 7.
Confusion matrix for multiple-vehicle accidents using various classifiers.
Fig 8.
Comparison of AUC–ROC of different machine learning classifiers.
(a) ROC curves for single-vehicles accidents. (b) ROC curves for multiple-vehicles accidents.