Metadata for cleaning, re-encoding, and documenting coded data columns.
These dictionaries are used to create Encoder instances. Each key is a table name with
a sub dictionary that includes additional detail. The table names must end with the
data_source as a sufix (for EIA 860, 861 or 923 tables include
The table-specific dictionaries contain the following keys:
‘df’: A dataframe associating short codes with long descriptions and other information. Each dataframe needs at least three standard columns: “code”, “label”, “description”. The codes and lables must be unique. By convention, the “label“‘s are snake case.
‘code_fixes’: A dictionary mapping non-standard codes to canonical, standardized codes.
‘ignored_codes’: A list of non-standard codes which appear in the data, and will be set to NA.