Apply Map for NaN rows in DataFrame, Python 3.6

Question

Topic_Details

                   Topic Source_Code      Anchor            Sub_Topic_Dataset            Dataset_Id
42  Macroeconomic Accounting Systems      GESAMT              Financial Accounts ESA 1995    DBB_GESAMTFAE12019
43  Macroeconomic Accounting Systems      GESAMT  GESAMTFINZ  Financial Accounts ESA 2010  DBB_GESAMTFINANZ2019
44  Macroeconomic Accounting Systems      GESAMT  GESAMTVOLK            National Accounts    DBB_GESAMTVOLK2019

Dataset_List:

       anchor                     text_eng            Dataset_Id
0  GESAMTVOLK            National             DBB_GESAMTVOLK2019
0  GESAMTFINZ  Financial accounts           DBB_GESAMTFINANZ2019
0              Financial accounts ESA 1995                   NaN

Map Script:

MapDF = dict(zip(Topic_Details['Sub_Topic_Dataset'].str.upper(), Topic_Details['Dataset_Id']))
Dataset_List['Dataset_Id'] = Dataset_List['text_eng'].apply(str.upper).map(MapDF)

I just want to run below script for the Dataset Dataset_List where Dataset_Id == NaN, not for other rows.

Dataset_List['Dataset_Id'] = Dataset_List['text_eng'].apply(str.upper).map(MapDF)

Answer 1

try passing the result of apply to fillna:

Dataset_List['Dataset_Id'].fillna(Dataset_List['text_eng'].apply(str.upper).map(MapDF), inplace=True)

It can accept a Series as a value, and for every NaN it uses the value with same index.