Source code for "FloodGNN-GRU: A Spatio-Temporal Graph Neural Network for Flood Prediction".
- Python == 3.8
- PyTorch == 1.11
- PyTorch Geometry == 2.0
- Scikit-learn == 1.1
- Scipy == 1.10
The data can be found here and should be placed in the 'data' folder (created by user) after download.
Details about the data can be found in the jupyter notebook data_exploration.ipynb
python train.py
Optional arguments:
--time_steps Total number of time steps.
--drop Dropout probability.
--batch_size Batch size.
--weight_decay
--n_epochs Number of epochs.
--lr Learing rate.
--offset Initial time step.
--v_h_dim Vector features - hidden dimensions.
--s_h_dim Scalar features - hidden dimensions.
Check Github repository for updates.