notebooks/
initial_experiments/ -- initial data exploration, first models
linear_models/ -- linear regression, ridge regression models
xgboost/ -- XGBoost models
neural_networks_non-spatial -- neural network models that don't explicitly use gridded data, but run on a flattened grid
neural_networks_convLSTM -- convolutional LSTM network that runs on gridded data
neural_networks_graph -- spatio-temporal graph-convolutional network that runs on graph of subbasin-relationships
visualization/ -- Visualizations of layers/kernels/...
evaluation/ -- Comparisons of physically-based and data-driven models
src/ -- common source code
conv_lstm.py -- code for convolutional LSTM models
datasets.py -- dataset classes to feed data into neural networks
evaluate.py -- code to evaluate prediction performance
load_data.py -- code to load and save data from/to disk
stgcn.py -- code for spatio-temporal graph-convolutional networks
utils.py -- helper functions
visualize.py -- visualization code
data/ -- used datasets (not in the repository for copyright and size reasons)
geophysical/ -- geophysical datasets
pickle/
models/ -- pickled trained models (not in the repository)
results/ -- pickled model results (not in the repository)
figures/ -- generated figures
requirements.txt -- python package requirements, to be installed with conda