Authors:
Carsten Baumann; DLR Institute for Solar-Terrestrial Physics, carsten.baumann@dlr.de
Aoife McCloskey; DLR Institute for Solar-Terrestrial Physics, aoife.mccloskey@dlr.de (alt:aoifemccloskey91@gmail.com)
This repository contains the program code for the Solar wind propagation delay machine learning approach using decision trees.
This repo includes:
the database (delay_DST_learningset.pickle) and code for:
stratification of database and k fold cross validation of the ML algorithms,
hyperparameter optimization of the ML algorithms,
comparison with other propagation delay models,
realtime scenario, (basic train/test approach but ordered in time),
drop column feature importance,
Shapley values to explain the models,
plotting routines
the repo for writing a manuscript is found under publications