[FIXED] How to drop inferior indexed rows if line is 0 across all superior index?

Issue

I have a multiindex dataframe in the form of :

level0	level1	1	2	3	4	5
idx1	name1	0.0	0.0	0.0	0.0	0.0
	name2	1.0	2.0	5.0	4.0	2.0
	name3	1.0	4.0	2.0	6.0	8.0
idx2	name1	0.0	0.0	0.0	0.0	0.0
	name2	4.0	2.0	9.0	5.0	5.0
	name3	0.0	0.0	0.0	0.0	0.0
idx3	name1	0.0	0.0	0.0	0.0	0.0
	name2	0.0	0.0	0.0	0.0	0.0
	name3	4.0	7.0	6.0	5.0	1.0

I would like to drop the rows where I have zeroes across all index of level 0. With the given exemple, (idx1, name1), (idx2, name1) and (idx3, name1) shoud be dropped.

I have tried : tmp.loc[~(tmp==0).all(axis=1)] But it drops all rows with zeroes...

I am not able to figure out the proper mask to get what I want. Can someone help ?

Solution

If I understand correctly, you further need to apply a groupby.transform to test homogeneity within a level:

tmp[(tmp!=0).any(axis=1).groupby(level='level1').transform('any')]

Which is identical to:

tmp[~(tmp==0).all(axis=1).groupby(level='level1').transform('all')]

Output:

                 1    2    3    4    5
level0 level1                         
idx1   name2   1.0  2.0  5.0  4.0  2.0
       name3   1.0  4.0  2.0  6.0  8.0
idx2   name2   4.0  2.0  9.0  5.0  5.0
       name3   0.0  0.0  0.0  0.0  0.0
idx3   name2   0.0  0.0  0.0  0.0  0.0
       name3   4.0  7.0  6.0  5.0  1.0

Intermediates:

                 1    2    3    4    5  (tmp!=0).any(axis=1)  (…).groupby(level='level1').transform('any')
level0 level1                                                                                             
idx1   name1   0.0  0.0  0.0  0.0  0.0                 False                                         False
       name2   1.0  2.0  5.0  4.0  2.0                  True                                          True
       name3   1.0  4.0  2.0  6.0  8.0                  True                                          True
idx2   name1   0.0  0.0  0.0  0.0  0.0                 False                                         False
       name2   4.0  2.0  9.0  5.0  5.0                  True                                          True
       name3   0.0  0.0  0.0  0.0  0.0                 False                                          True
idx3   name1   0.0  0.0  0.0  0.0  0.0                 False                                         False
       name2   0.0  0.0  0.0  0.0  0.0                 False                                          True
       name3   4.0  7.0  6.0  5.0  1.0                  True                                          True

Answered By - mozway