Need to re-tool cleaner logic using jupyter notebooks for prototyping and documenting:
previously used fields for some sites are obsolete (currently commented out)
parse duplicate fields before using it to fill missing data for base fields. This should lead to better complete data sets using most relevant values available in duplicate fields in raw data.
see #97: can improve groupset data integrity for missing values (this might actually have to be part of EDA data prep for model building)
Need to re-tool cleaner logic using jupyter notebooks for prototyping and documenting: