Sorting column by a groupby

Question

Hi all I have the following dataframe:

     a                 LC_REF
5 Warranty Tsunami      1C17
6 10.1                  1C17
7 5.2                   1C17
...
5 Warranty Medium       1C17
6 9.1                   1C17
7 6.3                   1C17
...
5 Warranty Medium       2C17
6 4.2                   2C17
7 3.1                   2C17
...
5 Warranty High         2C17 
6 2.1                   2C17
7 0.5                   2C17

I am trying to sort them by LC_REF, while maintaining the index order:

     a                 LC_REF
5 Warranty Medium       2C17
6 4.2                   2C17
7 3.1                   2C17
...
5 Warranty High         2C17
6 2.1                   2C17
7 0.5                   2C17
...
5 Warranty Tsunami      1C17
6 10.1                  1C17
7 5.2                   1C17
...
5 Warranty Medium       1C17 
6 9.1                   1C17
7 6.3                   1C17

However when I try:

df.sort_values('LC_REF') 

I get a very out of order dataframe. Is there any way to sort by LC_REF while still maintaining the order that they are in?

Answer 1

It seems you need:

df.sort_values(['LC_REF'], ascending=False)

More general solution - first reset_index for original index values and second reset_index for original order and then sort by 2 columns:

df = df.reset_index()
       .reset_index()
       .sort_values(['LC_REF', 'level_0'], ascending=[False, True])
       .set_index('index')
       .drop('level_0', 1)
print (df)
                      a LC_REF
index                         
5       Warranty Medium   2C17
6                   4.2   2C17
7                   3.1   2C17
5         Warranty High   2C17
6                   2.1   2C17
7                   0.5   2C17
5      Warranty Tsunami   1C17
6                  10.1   1C17
7                   5.2   1C17
5       Warranty Medium   1C17
6                   9.1   1C17
7                   6.3   1C17