Sort multiIndex dataframe given the sum-CodePudding

Hello I have prepared a MultiIndex table in Pandas that looks like this:

Lang                C    java  python  All
Corp     Name                             
ASW      ASW        0.0   7.0     8.0   15
         Cristiano  NaN   NaN     8.0   8
         Michael    NaN   7.0     0     7
Facebook Facebook   8.0   1.0     5.0   14
         Piter      8.0   NaN     NaN    8
         Cristiano  NaN   NaN     3.0    3
         Michael    NaN   1.0     2.0    3
Google   Google     2.0  24.0     1.0   27
         Michael    NaN  24.0     NaN   24
         Piter      2.0   NaN     NaN    2
         Cristiano  NaN   NaN     1.0    1

Now I would like to Sort group of rows where sum of Corp(in column "All') is sorted decsending, then I would like to select only the two index "Corp"(and their rows) which are the largest, It should looks like:

Lang                C    java  python  All
Corp     Name                             
Google   Google     2.0  24.0     1.0   27
         Michael    NaN  24.0     NaN   24
         Piter      2.0   NaN     NaN    2
         Cristiano  NaN   NaN     1.0    1
ASW      ASW        0.0   7.0     8.0   15
         Cristiano  NaN   NaN     8.0   8
         Michael    NaN   7.0     0     7

Thank You!

CodePudding user response：

IIUC, you can sort_values per group, then slice using the max sum of All per group:

out = (df
    # for each company, sort the values using the All column in descending order
   .groupby(level=0).apply(lambda g: g.sort_values('All', ascending=False))
    # calculate the sum or All per company
    # get the index of the top 2 companies (nlargest(2))
    # slice to keep only those
   .loc[lambda d: d.groupby(level=0)['All'].sum().nlargest(2).index]
)

output:

Lang                C    java  python  All
Corp     Name                             
Google   Google     2.0  24.0     1.0   27
         Michael    NaN  24.0     NaN   24
         Piter      2.0   NaN     NaN    2
         Cristiano  NaN   NaN     1.0    1
Facebook Facebook   8.0   1.0     5.0   14
         Piter      8.0   NaN     NaN    8
         Cristiano  NaN   NaN     3.0    3
         Michael    NaN   1.0     2.0    3