Replace zeros in a column with string from the row above (Python/Pandas)

Question

I would like to replace the 0 with the string from the same column, previous row. Eg: 0 under Sheffield should read Sheffield. I am working with pandas.

file = file[['Branch', 'Type' ,'total']]
#replace NaN with 0 
file.fillna(0).tail(6)
Out[48]: 
     Branch                     Type  total

394   Sheffield  Sum of Resend to Branch      0
395           0   Number of PV Enquiries     83
396   Wakefield  Sum of Resend to Branch      0
397           0   Number of PV Enquiries     38
398        York  Sum of Resend to Branch      1
399           0   Number of PV Enquiries     59

I have tried:
a) #create a  series for that column and replace
branch = file.iloc[ :, 0]
branch.replace(0, branch(-1))
# why is this series not callable?

b)# I tried a loop in the dataframe
for item in file:
    if "Branch" == 0:
        replace(0, "Branch"[-1])
# I am unsure how to refer to the row above

Answer 1

Use replace with the method ffill

file_df['Branch'].replace(to_replace='0', method='ffill', inplace=True)

>>> file_df
        Branch                     Type  total
394  Sheffield  Sum of Resend to Branch      0
395  Sheffield   Number of PV Enquiries     83
396  Wakefield  Sum of Resend to Branch      0
397  Wakefield   Number of PV Enquiries     38
398       York  Sum of Resend to Branch      1
399       York   Number of PV Enquiries     59

Or, since it looks like you already replaced the NaN with 0 , you could omit that step and just use ffill . ie if your original dataframe looks like:

>>> file_df
        Branch                     Type  total
394  Sheffield  Sum of Resend to Branch      0
395        NaN   Number of PV Enquiries     83
396  Wakefield  Sum of Resend to Branch      0
397        NaN   Number of PV Enquiries     38
398       York  Sum of Resend to Branch      1
399        NaN   Number of PV Enquiries     59

use:

file_df['Branch'].ffill(inplace=True)

Note that I called your dataframe file_df rather than file to not mask the python builtin