Intelligent Transportation Systems (ITS) leverage advanced technologies such as electronics, communications, and computer-based control systems, integrating them with traffic engineering and planning sciences to provide essential tools for traffic control and management. One of the most critical components of ITS is speed enforcement cameras. Among the most commonly used cameras in urban networks are Automatic Number Plate Recognition (ANPR) cameras.

This study examines traffic data recorded by two ANPR cameras: Camera No. 13, located at Shahid Kalantari Highway, East Gate of Ferdowsi University, Southeast of Mashhad, and Camera No. 14, positioned at Shahid Kalantari Highway, Seda va Sima Boulevard, Northwest of Mashhad. The dataset, collected during the first week of June 2020, includes vehicle license plate information, traffic violations, timestamps, vehicle speeds, and traffic volume.

These two cameras exhibited the highest error rates compared to other cameras in Mashhad. To identify and mitigate these errors, a Random Forest classification model was employed, analyzing various influencing parameters. The findings indicate that for license plate recognition errors, the most influential factor was the time of plate registration. The Random Forest model achieved an accuracy of approximately 85%, making it the most effective classifier for this error type.

In contrast, for odd/even plate violations, the lane in which the vehicle was traveling had the most significant impact on the likelihood of misclassification. The Random Forest model demonstrated an accuracy of 86% in detecting this type of violation