Anomaly detection in vehicle traffic flow plays a crucial role in ensuring efficient transportation systems and maintaining public safety. However, traditional methods for anomaly detection present certain limitations. For instance, older techniques often rely on statistical-based approaches, such as using standard deviation and assuming a normal data distribution, to identify anomalies based on statistical attributes. While these methods have paved the way for more advanced approaches, such as XGBoost and Isolation Forest, which capture complex patterns and relationships in the data, providing improved accuracy and flexibility in anomaly detection. This paper proposes a method for anomaly detection in vehicle traffic flow using XGBoost and Isolation Forest algorithms. XGBoost is a powerful gradient boosting framework that effectively captures complex patterns in the data, while Isolation Forest is an unsupervised learning algorithm that isolates anomalies based on their unique characteristics. The approach involves preprocessing the traffic data, extracting relevant features, and training the models using XGBoost and Isolation Forest. Experimental results on real-world traffic datasets demonstrate the effectiveness of the proposed method, achieving a high accuracy using a threshold of 75% for XGBoost and 40% for Isolation Forest in detecting anomalies. This approach has the potential to enhance traffic management systems and improve overall traffic flow efficiency.
Empowering Traffic Management: Anomaly Detection in Vehicle Traffic Flow Using XGBoost and Isolation Forest Algorithms
Lect. Notes Electrical Eng.
International Conference on Electrical, Control & Computer Engineering ; 2023 ; Kuala Lumpur, Malaysia August 22, 2023 - August 22, 2023
Proceedings of the 7th International Conference on Electrical, Control and Computer Engineering—Volume 2 ; Chapter : 30 ; 345-357
2024-09-15
13 pages
Article/Chapter (Book)
Electronic Resource
English
A Traffic Anomaly Detection Method Using Traffic Flow Vectors During Heavy Rainfall
Springer Verlag | 2025
|A Traffic Anomaly Detection Method Using Traffic Flow Vectors During Heavy Rainfall
Springer Verlag | 2025
|European Patent Office | 2023
|Taylor & Francis Verlag | 2025
|