Jurnal Institusi
Copyright (c) 2025 ComniTech : Journal of Computational Intelligence and Informatics
Traffic congestion in Pekanbaru has become a critical urban challenge due to rapid growth in vehicle volume that is not matched by corresponding road capacity expansion. This study aims to develop a classification model that categorizes traffic congestion levels using original operational data obtained from the local Transportation Agency. The proposed approach employs a Random Forest classifier with three primary features: speed, traffic density, and the volume to capacity (V/C) ratio. Congestion levels are defined in three classes, low, moderate, and high, based on rule based thresholds derived from the Highway Capacity Manual (HCM) traffic engineering standards. The results demonstrate that the developed model achieves excellent predictive performance, reaching 98.86% accuracy on the testing dataset. Feature importance analysis further confirms that speed and the V/C ratio are the most dominant variables in determining congestion severity. Overall, the model is effective as a decision support tool for rapid identification of congestion hotspots and provides a practical foundation for developing a more responsive transportation management system in Pekanbaru
ComniTech : Journal of Computational Intelligence and Informatics ; Vol. 2 No. 2 (2025): ComniTech : Journal of Computational Intelligence and Informatics; 80-89
Penerbit: Universitas Lancang Kuning