Name		Name	Last commit message	Last commit date
parent directory ..
ganmodels		ganmodels
logs/2.GAN-training		logs/2.GAN-training
1.Compression.ipynb		1.Compression.ipynb
2.GAN-training.ipynb		2.GAN-training.ipynb
3.GAN-Prediction.ipynb		3.GAN-Prediction.ipynb
GAN_0-end_p4.png		GAN_0-end_p4.png
GAN_0-end_p84.png		GAN_0-end_p84.png
GAN_2000-end_p4.png		GAN_2000-end_p4.png
GAN_2000-end_p84.png		GAN_2000-end_p84.png
README.md		README.md
X_generated2000.npy		X_generated2000.npy
X_generatedall.npy		X_generatedall.npy
X_train_1D.csv		X_train_1D.csv
X_train_encoded.csv		X_train_encoded.csv
X_train_pca.csv		X_train_pca.csv
conv_ae.h5		conv_ae.h5
environment.yml		environment.yml
gan-tfex.h5		gan-tfex.h5
group-output-time_mod.csv		group-output-time_mod.csv
groups.txt		groups.txt
meshprgan_2000.png		meshprgan_2000.png
meshprgan_2000_diff.png		meshprgan_2000_diff.png
pca_compress_15.pkl		pca_compress_15.pkl
scaler.pkl		scaler.pkl
stdscaler.pkl		stdscaler.pkl
times.csv		times.csv

README.md

(outdated see DA-PredGAN/UQ-PredGAN folder) Predictive GAN (PredGAN) - Forecasting spatial variation of COVID-19 infection using GAN

1.Compression.ipynb -> Compress the 800 variables per time step to 15 variables per time step (the compression used for training and prediction is from PCA. However, there are in this file other types of compression using autoencoders).
2.GAN-training.ipynb -> Train a GAN and save the model
3.GAN-Prediction.ipynb -> Predict with the GAN

To execute:

From inside the notebooks

Cell->Run All

From the command line

$ jupyter nbconvert --to notebook --execute <notebookname>.ipynb

Requirements

To install requirements:

 $ conda env create -f environment.yml 
 $ conda activate py3ml
 $ python -m ipykernel install --user --name=python3 (optional)

Finally, start Jupyter:

 $ jupyter notebook