STXGCN

R Libraries required: evmix, forecast, dfms, gstar, xts

Python libraries required are listed in the requirements.txt file.

Folders

data contains the cleaned data of the air pollutants along with their locations and the Haversine matrix.

rscripts conatins all the R codes that have been implemented for the model.

pyscripts conatins all the python codes that have been implemented for the model.

results contains the folders to the results for all the models.

Obtaining the results

To get the results for Peaks over threshold run Fitting_GPD.R in the rscripts folder. Change the chunk of code for the data_folder to the location where you have cloned the repository.

data_folder <- "C:/Users/anubhab.biswas/OneDrive - SUPSI/Documenti/"

Also set a new storage folder for saving the results by changing the following code.

result_folder <- paste0(data_folder,"STXGCN/results/")

Note that a subfolder called pot_results inside result_folder will be required.

For obtaining the results for individual baseline models create the following sub-folders inside result_folder:

ha, arima, dfm, gstar, lstm, stgcn, stxgcn

Model Architecture