Batch Transform Data

If you don’t have a preference or idea for features, just try with BatchFeature , We using features from pymatgen firstly.

  • Transform structure list.

>>> from featurebox.featurizers.batch_feature import BatchFeature
>>> bf = BatchFeature(data_type="structures", return_type="df")
>>> data = bf.fit_transform(structure_list)

structures_list is list of struceture of pymatgen.

../_images/structures0.gif

  • Transform composition list.

>>> from featurebox.featurizers.batch_feature import BatchFeature
>>> bf = BatchFeature(data_type="compositions")
>>> com = [[{str(i.symbol): 1} for i in structurei.species]  for structurei in structure_list]
>>> #where com is element list
>>> data = bf.fit_transform(com)
../_images/composition0.gif

  • Transform element list.

>>> from featurebox.featurizers.batch_feature import BatchFeature
>>> bf = BatchFeature(data_type="elements")
>>> aas = [[{str(i.symbol): 1} for i in structurei.species]  for structurei in structure_list]
>>> data = bf.fit_transform(aas)
>>> bf.element_c.search_tp="number"
>>> aas = [[i.specie.Z for i in structure] for structure in structure_list]
>>> data = bf.fit_transform(aas)

Note

It is highly recommended that using this function as a beginner,
Because we can customize more and more powerful converters.

Now, try it !