Data processing for the JuHemd database and ML-training and evaluation scripts

This software publication contains python code which: 

1. Process the JuHemd (see related identifiers) database so that complete data points are selected to be processed in a machine readable and ML-ready format. A magnetic threshold of 0.1 Bohr magneton as total absolute magnetic moment is applied to the data. Data points with zero critical temperatures are excluded. The script produces data arrays in randomized order and uses supplemental atomic data. The supplemental atomic data as well as the generated data set is published separately (see related identifiers). The data is generated once including density-functional theory generated descriptors and once excluding them. 

2. Is needed to train and evaluate ML-models demonstrating the use of the data for the prediction of critical temperatures as described in our upcoming paper.