CrystalFeatures is a set of python codes that extract features of crystal systems, using only the CIF file as input, for machine learning applications. There are three classes of features in CrystalFeatures: basic descriptors, which include statistical properties of the atoms in the materials as well as the crystal symmetry; geometry features, which are based on bonding properties; and SDFT features, which stands for "superficial" DFT, where the eigen values of the material are obtained using a single-point single step calculation is performed using GPAW.
We are submitting this package for peer review soon.