Sort pandas df subset of rows (within a group) by specific column

Question

I have the following dataframe let's say:

df

A B C D E
z k s 7 d
z k s 6 l
x t r 2 e
x t r 1 x
u c r 8 f
u c r 9 h
y t s 5 l
y t s 2 o

And I would like to sort it based on col D for each sub row (that has for example same cols A,B and C in this case)

The expected output would be:

df

A B C D E
z k s 6 l
z k s 7 d
x t r 1 x
x t r 2 e
u c r 8 f
u c r 9 h
y t s 2 o
y t s 5 l

Any help for this kind of operation?

Answer 1

I think it should be as simple as this:

df = df.sort_values(["A", "B", "C", "D"])

Answer 2

You can use groupby and sort values (also credit to @Henry Ecker for his comment):

df.groupby(['A','B','C'],group_keys=False,sort=False).apply(pd.DataFrame.sort_values,'D')

output:

    A   B   C   D   E
1   z   k   s   6   l
0   z   k   s   7   d
3   x   t   r   1   x
2   x   t   r   2   e
4   u   c   r   8   f
5   u   c   r   9   h
7   y   t   s   2   o
6   y   t   s   5   l

Answer 3

Let us try ngroup create the help col

df['new1'] = df.groupby(['A','B','C'],sort=False).ngroup()
df = df.sort_values(['new1','D']).drop('new1',axis=1)
df
   A  B  C  D  E
1  z  k  s  6  l
0  z  k  s  7  d
3  x  t  r  1  x
2  x  t  r  2  e
4  u  c  r  8  f
5  u  c  r  9  h
7  y  t  s  2  o
6  y  t  s  5  l

Answer 4

dic = {
    'A': [*'zzxxuuyy'],
    'B': [*'kkttcctt'],
    'C': [*'ssrrrrss'],
    'D': [*map(int, '76218952')],
    'E': [*'dlexfhlo']
}
df = pd.DataFrame(dic)

df.groupby(['A', 'B']).apply(lambda df: df.sort_values('D')).droplevel(['A', 'B']).reset_index()

if you want to sort based on columns 'A', 'B', 'C', 'E' then you have to:

df.groupby(['A', 'B', 'D', 'E']).apply(lambda df: df.sort_values('D')).droplevel(['A', 'B', 'D', 'E']).reset_index()