MG-TAR: Multi-view Graph Convolutional Networks for Traffic Accident Risk Prediction

This is the implementation of a paper published in IEEE Transactions on Intelligent Transportation Systems (Volume: 24 Issue: 4) [Paper]

Citation

@article{trirat2023mgtar,
  author={Trirat, Patara and Yoon, Susik and Lee, Jae-Gil},
  journal={IEEE Transactions on Intelligent Transportation Systems}, 
  title={{MG-TAR}: Multi-View Graph Convolutional Networks for Traffic Accident Risk Prediction}, 
  year={2023},
  volume={24}, 
  number={4},
  pages={3779-3794},
  doi={10.1109/TITS.2023.3237072}}
}   

Abstract

Due to the continuing colossal socio-economic losses caused by traffic accidents, it is of prime importance to precisely forecast the traffic accident risk to reduce future accidents. In this paper, we use dangerous driving statistics from driving log data and multi-graph learning to enhance predictive performance. We first conduct geographical and temporal correlation analyses to quantify the relationship between dangerous driving and actual accidents. Then, to learn various dependencies between districts besides the traditional adjacency matrix, we simultaneously model both static and dynamic graphs representing the spatio-temporal contextual relationships with heterogeneous environmental data, including the dangerous driving behavior. A graph is generated for each type of the relationships. Ultimately, we propose an end-to-end framework, called MG-TAR, to effectively learn the association of multiple graphs for accident risk prediction by adopting multi-view graph neural networks with a multi-attention module. Thorough experiments on ten real-world datasets show that, compared with state-of-the-art methods, MG-TAR reduces the error of predicting the accident risk by up to 23% and improves the accuracy of predicting the most dangerous areas by up to 27%.

Example Run

• Package Installation: pip install -r requirements.txt
• (Contextual) Graph Preprocessing: multi-view_graph_construction.ipynb
• MG-TAR Model Train-Test Demo: example_run.ipynb

Note for Driving Record Data

• Digital Tachograph (Driving Log) Data: cannot be publicly accessible due to non-disclosure agreements

  • For demonstration purpose, we partially provide the aggregated number of classified dangerous driving cases in the datasets folder.
  • If you are interested in the original data, there is a sample file provided here by Korea Transportation Safety Authority.

• Acknowledgement