Each script takes parameters in input: read the file Command.txt to see the commands to run the script. Each folder contains a Command.txt file in which there are the ordered instructions.
You can find the DataSet folder at the link https://drive.google.com/drive/folders/1y3xJH6ZsLRlkzL7mQv7QHy-arDWGE2uO?usp=sharing. The data created with Fluidity are serialized and stored in the file 'data' at the paths DataSet/Structured/ and DataSet/Structured/ for the Structured and RGB datasets respectively. At the same path are stored also the normalized data in the file 'scaled' and the train and test dataset.
The DataSet folder contains both dataset only. Assuming you want to work with the Structured dataset, you should normalize the data using the code in NormalizePPM.py and you should create the train and test sets trough the train_test_split.py code.
All the configurations of the Autoencoder tested are stored at the path AutoEncoder/Structured/LS7/. For example, we first analyzed how many Convolutional Layers the autoencoder should have. In the file AutoEncoder/Structured/LS7/ExpLayers.py the are three different autoencoder structures and it's specified the path where the results must be stored also (in this case, the results are stored in the folder AutoEncoder/Structured/LS7/ResultLayers). We used the script AutoEncoder/Experiments.py to test the different configurations. Finally, for the Grid Search we used the script GridSearchAE.py The results are analyzed in the notebook AutoEncoder/AnalysisLS7.ipynb. The trained Autoencoder is stored at the path AutoEncoder/Structured/LS7/model-64-relu-32-400.h5.
The train and validation set for the LSTM are created with the script train_val.py and they are stored at the path LSTM/Structured/. We tested different configurations of LSTM with the script LSTM/LSTM_Layers.py and we perform the Grid Search with the script LSTM/LSTM_GS.py. All results are stored at the path LSTM/Structured/ and they are analyezed in the notebook AutoEncoder/AnalysisLS7.ipynb. The final LSTM is trained through the script fit_lstm.py and it is stored at the path LSTM/Structured/model-3-elu-30-400-16.h5.
The data of the observations are stored at the path DataAssimilation/SensorsData/Observation.xlsx. The script LatentAssimilation.py performs the assimilation in the Latent Space and it prints the table of the results.