<![CDATA[The Traffic accidents are a complex issue influenced by multiple factors, including human behavior, vehicle conditions, road characteristics, and environmental conditions. Historically, traffic accident data maintained by the Police Department of Bengkulu City have been utilized primarily for descriptive reporting, limiting their potential to reveal comprehensive patterns of accident causation. This study aims to identify the dominant factors contributing to traffic accidents, implement the Decision Tree C4.5 algorithm for accident classification, and evaluate the performance of the resulting classification model. This study used the Knowledge Discovery in Databases (KDD) methodology. The dataset consisted of historical traffic accident records obtained from the Jurisdiction of Bengkulu City Police. Data processing was conducted using Python in the Google Collaboratory environment through several stages, including exploratory data analysis (EDA), data preprocessing, model development, and performance evaluation. The results demonstrate that the proposed model successfully identified relationships among accident causative factors through the construction of a decision tree, generating 20 classification rules. Road Surface Condition was identified as the most influential predictor, with a feature importance value of 0.30, followed by Special Information (0.15), Speed Limit at the Accident Location (0.13), and Lighting Condition (0.10). Model evaluation showed an overall accuracy of 65.2%, with a weighted average F1-score of 75.9%, indicating satisfactory performance in classifying the majority classes. However, the macro average F1-score of 28.9% suggests that the model's performance on minority classes remains limited due to class imbalance in the dataset. This study is expected to provide a data-driven foundation for developing more effective traffic accident prevention and mitigation strategies in the Jurisdiction of Bengkulu City Police.]]>
Copyrights © 2026
