Table 10.

Literature result (continued, Part 5 of 5): summary table of papers reporting on objective, analysis methods, and obtained results.

Author(s)	Objective	Analysis Methods	Results
Vlahogianni and Barmpounakis, 2017 [108]	Detect driving events such as braking, acceleration, left and right cornering	Rough set theory and own classifier (MODLEM), compared to MLP, C4.5 decision trees, and ZeroR	Smartphone accuracy is 99.4% and OBD-II device accuracy is 99.3%; TPRs are 88% and 86% and FPRs are 0.3% and 0.4% for smartphone and OBD-II device, respectively,
Woo and Kulic, 2016 [109]	Propose a classifier-based approach for driving manoeuvre recognition from mobile phone data	SVM classifier, PCA	Average precision of 0.8158 and average recall of 82%. Balanced accuracy of 88%.
Xiao and Feng, 2016 [111]	Driver attention detection with 2 modules: a) gaze detection and b) road motion objects detection	Linear SVM classifier (module a); Lucas–Kanade optical flow with dynamic background compensation (module b)	93% accuracy for gaze estimation and 91.7% overall accuracy
Xie and Zhu, 2019 [110]	Manoeuvre-based driving behaviour (lane changing or turning) and classification amongst three labels (normal, drowsy, and aggressive)	ReliefF, random forest	Average F1 score of 70.47% using leave-one-driver-out validation
Xie et al., 2018 [112]	Classification of driving manoeuvres (i.e., braking, turning, stopping, accelerating, decelerating, lane changing) based on different feature extraction methods	Random forest classifier	F1 scores of 68%, 80%, and 87% on three different datasets
Xie et al., 2019 [113]	Driver distraction detection	Ensemble method of 4 classifiers: K-NN, Logistic Regression, Gaussian Naive Bayes, random forest	87% accuracy in distraction detection
Yang et al., 2012 [114]	Distinguish between passengers and drivers using smartphones by classifying the position of the smartphone	Threshold-based classification	Accuracy with calibrated thresholds: detection rate is over 90% and accuracy is around 95%
Yaswanth et al., 2021 [115]	Smartphone detection (classifier) and drivers’ action detection	N/A	N/A
You et al., 2013 [116]	Detect if drivers are tired or distracted (drowsy driving, inattentive driving) and identify various driving conditions such as tailgating, lane weaving, or drifting	Computer vision and machine learning (decision trees and SVM)	Precision and recall for face direction events: precisions are 68% for facing left, 79% for facing right, and 92% for eye state classification
Ziakopoulos et al., 2023 [117]	Investigate influence factors for driver distraction through smartphone use	230-driver experiment using the developed driving recording application and feedback questionnaire, XGBoost for distraction investigation	Deducted influence factors for driver phone use