Use pandas.Categorical for enum types

[sylvlecl]

• Do you want to request a feature or report a bug?

Feature

• What is the current behavior?

Enum attributes are represented as strings in the provided dataframes, which is not memory efficient and not very descriptive.

• What is the expected behavior?

pandas proposes the pd.Categorical type for objects with only a few valid values.

• What is the motivation / use case for changing the behavior?

Improved memory efficiency and semantics.

About memory efficiency, see on a large network:

>>> network.get_generators().memory_usage(deep=True)
...
energy_source           316067
...

>>> network.get_generators().astype('category').memory_usage(deep=True)
...
energy_source             5652

We can also improve the java -> python performances by transferring "codes" (integers) instead of names, and create the series from the underlying codes:

series = pd.Series(pd.Categorical.from_codes(data, dtype=EnergySource))

[sylvlecl]

Note that this will probably require to add some information about the enum type in the series metadata, in order to know which categorical type to use on python side.