Tukukkk

I have a pandas dataframe like this:

df = pd.DataFrame({"RT":[9,10,10,11,11,11,11],"Quality":[70,60,50,60,80,70,80],'Name' :['a','a','b','c','b','c','b'],'Similarity':[0.98,0.97,0.97,0.95,0.95,0.95,0.95]})



    RT  Quality Name    Similarity

0   9   70      a       0.98

1   10  60      a       0.97

2   10  50      b       0.97

3   11  60      c       0.95

4   11  80      b       0.95

5   11  70      c       0.95

6   11  80      b       0.95

The values in the column Similarity has the same group-by with column RT

I want to group column RT and find the maximum column Quality value and group by column Name.

For example:

In column RT value 11,which have column Name value c and b, sum each of the column Quality values, then get c = 130, b =160, and sort the maximum 160, b
then get

    RT  Quality Name    Similarity

0   9   70  a       0.98

1   10  60  a       0.97

2   10  50  b       0.97

3   11  160 b       0.95

4   11  130 c       0.95

you can use groupby with agg:

use lambda to return all Similarities or max to return the max

df.groupby(['RT','Name']).agg({'Quality':'sum', 'Similarity':lambda x:x.unique()})



         Quality    Similarity

RT  Name        

9   a     70        0.98

10  a     60        0.97

    b     50        0.97

11  b     160       0.95

    c     130       0.95

You may not need agg

df.groupby(['RT','Similarity','Name'],as_index=False)['Quality'].sum()

Out[150]: 

   RT  Similarity Name  Quality

0   9        0.98    a       70

1  10        0.97    a       60

2  10        0.97    b       50

3  11        0.95    b      160

4  11        0.95    c      130