I have two dataframes ( df1 & df2) with same headings.
Both dataframes contains same number of rows.
There are around 20 columns in each dataframes.
The dataframes differs for some columns (any one or more out of 20 columns)
There is a particular column (ssno) which is unique for both.
I need to generate the output(ssno) for those rows which differes in any of the 20 field'.
Please help.
First compare both DataFrames and get al least one Trues per rows by any and then use boolean indexing for filtering ssno column:
df1 = pd.DataFrame({'ssno':list('abcdef'),
'B':[4,5,4,5,5,4],
'C':[70,8,9,4,2,3],
'D':[1,3,5,7,1,0],
'E':[5,3,6,90,2,4],
'F':list('aaXbbb')})
print (df1)
B C D E F ssno
0 4 70 1 5 a a
1 5 8 3 3 a b
2 4 9 5 6 X c
3 5 4 7 90 b d
4 5 2 1 2 b e
5 4 3 0 4 b f
df2 = pd.DataFrame({'ssno':list('abcdef'),
'B':[4,5,4,5,5,4],
'C':[7,8,9,4,2,3],
'D':[1,3,5,7,1,0],
'E':[5,3,6,9,2,4],
'F':list('aaabbb')})
print (df2)
B C D E F ssno
0 4 7 1 5 a a
1 5 8 3 3 a b
2 4 9 5 6 a c
3 5 4 7 9 b d
4 5 2 1 2 b e
s = df1.loc[(df1 != df2).any(1), 'ssno']
print (s)
0 a
2 c
3 d
Name: ssno, dtype: object
Detail:
print (df1 != df2)
B C D E F ssno
0 False True False False False False
1 False False False False False False
2 False False False False True False
3 False False False True False False
4 False False False False False False
5 False False False False False False
print ((df1 != df2).any(1))
0 True
1 False
2 True
3 True
4 False
5 False
dtype: bool
s2 = df1.columns[(df1 != df2).any()] ?s2 = df1[(df1 != df2)].stack().rename_axis(('idx','cols')).reset_index(name='val_df1') ?