Table 6 Comparison of machine learning models for accident prediction.
Study | Model | Accuracy (%) | Key Features | Strengths | Limitations | |
|---|---|---|---|---|---|---|
Aboulola66 | MobileNet | 98.17 | Deep learning, CNN-based | High accuracy | Computationally expensive | |
| Â | Decision Tree | 85.4 | Feature-based classification | Interpretability | Prone to overfitting | |
Ardakani et al.67 | Random Forest | 87.6 | Ensemble learning approach | Robust performance | Slower inference time | |
|  | Naïve Bayes | 60.0 | Probabilistic classification | Fast computation | Low prediction power | |
Kim et al.69 | Neural Network | 94.59 | Gradient boosting, DTG data | High predictive power | Require large dataset | |
Khosravi et al.70 | K-Nearest Neighbors (KNN) | 71.0 | Distance-based classification | Simple to implement | Lower accuracy | |
Random Forest | 60.0 | Road & environmental attributes | Identifies accident hotspots | Limited dataset size | ||
