Calculating difference between two pandas dataframes with same columns and some mismatching rows

Question

df1

fileName     obj1   obj2   obj3   obj4
file_01.jpg  1      1      1
file_02.jpg         1      1
file_03.jpg  1      2             1

df2

fileName     obj1   obj2   obj3   obj4
file_01.jpg  1             2
file_02.jpg         1      1
file_04.jpg         3      1      2

Say I have two pandas DataFrames like above, question: how can I calculate the difference between te two to get a final DataFrame like below?

Expected output
resultdf

fileName     obj1   obj2   obj3   obj4
file_01.jpg  0      1      -1      0
file_02.jpg  0      0      0      0
file_03.jpg  1      2      0      1      
file_04.jpg  0      -3     -1     -2

What I've already tried : I reached the result by merging the two DataFrames together, and then calculating the difference between two columns. Raising this question in SO to see if there is any other efficient method.

Answer 1

I can't seem to find a groupby subtract so I multiplied your second df by -1 so sum could product the desired result

obj_cols = ['obj1', 'obj2', 'obj3', 'obj4']
df2[obj_cols] *= -1

pd.concat([df1, df2]).groupby(['fileName'], as_index=False).sum()

      fileName  obj1  obj2  obj3  obj4
0  file_01.jpg     0     1    -1     0
1  file_02.jpg     0     0     0     0
2  file_03.jpg     1     2     0     1
3  file_04.jpg     0    -3    -1    -2