W-DSTAGNN

Wavelet-based Temporal Attention Improves Traffic Forecasting

Requirements

• python >= 3.5
• scipy
• tensorboardX
• pytorch
• scikit-learn
• tqdm

Datasets

W-DSTAGNN is implemented on these public traffic datasets.

• PEMSBAY from DCRNN (ICLR-18)
• PEMS03, PEMS04 from STSGCN (AAAI-20) All datasets can be found in .npz format with the only key being 'data' in Google drive. The shape of input traffic data should be (Total_Time_Steps, Node_Number).

Processing the datasets:

• on PEMSBAY dataset

  python prepareData.py --config configurations/PEMSBAY_dstagnn.conf

• on PEMS03 dataset

  python prepareData.py --config configurations/PEMS03_dstagnn.conf

• on PEMS04 dataset

  python prepareData.py --config configurations/PEMS04_dstagnn.conf

Spatial-Temporal Aware Graph Construction

The graphs can be generated by code:

cd ./data/
python STAG_gen.py

The calculation uses CPU, which should be prepared for enough computation resources.

Train and Test

• on PEMSBAY dataset

  python train_DSTAGNN.py --config configurations/PEMSBAY_dstagnn.conf   

• on PEMS03 dataset

  python train_DSTAGNN.py --config configurations/PEMS03_dstagnn.conf   

• on PEMS04 dataset

  python train_DSTAGNN.py --config configurations/PEMS04_dstagnn.conf   

Configuration

The configuration file config.conf contains two parts: Data, Training:

Data

• adj_filename: path of the adjacency matrix file
• graph_signal_matrix_filename: path of graph signal matrix file
• stag_filename: path of the Spatial-Temporal Aware Grap file
• strg_filename: path of the Spatial-Temporal Relevance Graph file
• num_of_vertices: number of vertices
• points_per_hour: points per hour, in our dataset is 12
• num_for_predict: points to predict, in our model is 12
• len_input: length of each input to the model, in our model is 12
• dataset_name: name of the dataset

Training

• graph: select the graph structure, G or AG, G stands for adjacency graph, AG stands for Spatial-Temporal Aware Graph
• ctx: set ctx = cpu, or set gpu-0, which means the first gpu device
• in_channels: number of channels for spatial convolution
• n_heads: int, number of temporal att heads will be used
• K: int, K-order chebyshev polynomials (number of spatial att heads) will be used
• d_k: int, the dimensions of the Q, K, and V vectors will be used
• d_model: int, d_E
• batch_size: int
• model_name: name of the model
• num_of_weeks: int, how many weeks' data will be used
• num_of_days: int, how many days' data will be used
• num_of_hours: int, how many hours' data will be used
• epochs: int
• learning_rate: float

Citation

If you find this repository useful in your research, please cite the following paper:

@inproceedings{wdstagnn,
  title={Wavelet-based Temporal Attention Improves Traffic Forecasting},
  author={Jakhmola, Yash and Kumar Mishra, Nitish and Ghosh, Kripabandhu and Chakraborty, Tanujit},
  year={2024}
}