使用groupby和mean()在Pandas中保留一个带有分类变量的列

Question

有没有办法保持分类变量之后groupby和mean()？例如,给定数据帧df:

              ratio    Metadata_A      Metadata_B   treatment
0      54265.937500           B10               1  AB_cmpd_01
11    107364.750000           B10               2  AB_cmpd_01
22     95766.500000           B10               3  AB_cmpd_01
24     64346.250000           B10               4  AB_cmpd_01
25     52726.333333           B10               5  AB_cmpd_01
30     65056.600000           B11               1          UT
41     78409.600000           B11               2          UT
52    133533.000000           B11               3          UT
54    102433.571429           B11               4          UT
55     82217.588235           B11               5          UT
60     89843.600000            B2               1          UT
71     98544.000000            B2               2          UT
82    179330.000000            B2               3          UT
84    107132.400000            B2               4          UT
85     73096.909091            B2               5          UT

我需要ratio在每个内部进行平均Metadata_A,但最后要保留列treatment:

从理论上讲,类似于:

df.groupby(by='Metadata_A').mean().reset_index()

              ratio    Metadata_A      Metadata_B   treatment
 0     54265.937500           B10             2.5  AB_cmpd_01
 1     78409.600000           B11             2.5          UT
 2    107132.400000            B2             2.5          UT

但是,treatment平均后列会消失.

Answer 1

你可以用groupby与agg

df.groupby(['Metadata_A','treatment'],as_index=False).agg({'Metadata_B':'mean','ratio':'first'})
Out[358]: 
  Metadata_A   treatment  Metadata_B       ratio
0        B10  AB_cmpd_01           3  54265.9375
1        B11          UT           3  65056.6000
2         B2          UT           3  89843.6000