Select MultiIndex rows by level, in Pandas-CodePudding

How can I select rows from a MultiIndex DataFrame that have more than 1 level? For example, given the following DataFrame:

           col
L1  L2     
a   1      5624
    2      1656
    3      265677
    4      3755
b   5      47
    6      85544
c   7      97656
d   8      12774
e   9      111
    10     9478

I would like to end up with a DataFrame that looks like:

         col
one two     
a   1      5624
    2      1656
    3      265677
    4      3755
b   5      47
    6      85544
e   9      111
    10     9478

CodePudding user response：

Check transform count

out = df[df.groupby(level=0)['col'].transform('count').values>1]

CodePudding user response：

If you only want to consider the first level:

df[df.index.get_level_values('L1').duplicated(keep=False)]
# or
df[df.index.get_level_values(0).duplicated(keep=False)]

If you want to consider all levels:

df[df.index.to_frame().apply(pd.Series.duplicated, keep=False).any(1)]

output:

          col
L1 L2        
a  1     5624
   2     1656
   3   265677
   4     3755
b  5       47
   6    85544
e  9      111
   10    9478