使用Pandas在数据透视表上运行groupby

Question

I have a pivot table that looks like this:

In [41]: counts
Out[41]: 
SourceColumnID                    3029903181  3029903182  3029903183  3029903184  ResponseCount
ColID      QuestionID RowID                                                                    
3029903193 316923119  3029903189         773         788         778         803           3142
3029903194 316923119  3029903189         766         799         782         773           3120

[2 rows x 5 columns]

and I'm trying to figure out how I can groupby RowID so that I can get total counts for each column for each RowID (in this case it would just sum up all of them since the 2 are in the same rowid).

This is the pivot tables index:

In [42]: counts.index
Out[42]: 
MultiIndex(levels=[[3029903193, 3029903194], [316923119], [3029903189]],
           labels=[[0, 1], [0, 0], [0, 0]],
           names=[u'ColID', u'QuestionID', u'RowID'])

Answer 1

You'll want to groupby 'RowID' . Since it's a level on the MultiIndex you pass 'RowID' to the level keyword.

In [5]: df.groupby(level='RowID').sum()
Out[5]: 
            3029903181  3029903182  3029903183  3029903184  ResponseCount
RowID                                                                    
3029903189        1539        1587        1560        1576           6262