richomelbds / graph_convolutional_lstm Goto Github PK

• The 2nd version of the structure of Traffic Graph Convolutional LSTM (TGC-LSTM).
  • is the K-th order adjacency matrix
  • is the Free Flow Reachability matrix defined based on the network physical topology information.
• The traffic graph convolution module is designed based on the physical network topology.
• The code of this model is in the Code_V2 folder.
  • Environment (Jupyter Notebook): Python 3.6.1 and PyTorch 0.4.1
  • The code contains the implementations and results of the compared models, including LSTM, spectral graph convolution LSTM, localized spectral graph convolution LSTM.

• The 1st version of Traffic Graph Convolutional LSTM.
• The code of this model is in the Code_V1 folder.
  • Environment: Python 3.6.1 and PyTorch 0.3.0

The model is tested on two real-world network-wide traffic speed dataset, loop detector data and INRIX data. The following figure shows the covered areas. (a) Seattle freeway network; (b) Seattle downtown roadway network.

Check out this Link for looking into and downloading the loop detecotr dataset. For confidentiality reasons, the INRIX dataset can not be shared.

To run the code, you need to download the loop detector data and the network topology information and put them in the proper "Data" folder.

For more detailed experimental results, please refer to the paper.

Please cite our paper if you use this code or data in your own work: Traffic Graph Convolutional Recurrent Neural Network: A Deep Learning Framework for Network-Scale Traffic Learning and Forecasting

Hope our work is benefitial for you. Thanks!

@article{cui2019traffic,
  title={Traffic graph convolutional recurrent neural network: A deep learning framework for network-scale traffic learning and forecasting},
  author={Cui, Zhiyong and Henrickson, Kristian and Ke, Ruimin and Wang, Yinhai},
  journal={IEEE Transactions on Intelligent Transportation Systems},
  year={2019},
  publisher={IEEE}
}