DataFrame.groupby(by=None, axis=0, level=None, as_index=True, sort=True, group_keys=True, squeeze=False, observed=False, **kwargs

> import pandas as pd
> df = pd.DataFrame({'A': ['a', 'b', 'a', 'c', 'a', 'c', 'b', 'c'], 'B': [2, 8, 1, 4,
3, 2, 5, 9], 'C': [102, 98, 107, 104, 115, 87, 92, 123], 'D': [2, 98, 17, 14, 15, 7, 92,
 13]})
> df
  A B  C  D
0 a 2 102  2
1 b 8  98 98
2 a 1 107 17
3 c 4 104 14
4 a 3 115 15
5 c 2  87  7
6 b 5  92 92
7 c 9 123 13
> df.groupby(by='A').sum()
  B  C  D
A
a  6 324  34
b 13 190 190
c 15 314  34

> df.groupby(by=['A','B']).sum()       ###A,B成索引
    C  D
A B
a 1 107 17
 2 102  2
 3 115 15
b 5  92 92
 8  98 98
c 2  87  7
 4 104 14
 9 123 13

> df.groupby(by=['A','B'])['C'].sum()   ###1个列
A B
a 1  107
  2  102
  3  115
b 5   92
  8   98
c 2   87
  4  104
  9  123
 
> df.groupby(by=['A','B'])['C','D'].sum()  ###2个列
    C  D
A B
a 1 107 17
 2 102  2
 3 115 15
b 5  92 92
 8  98 98
c 2  87  7
 4 104 14
 9 123 13

> import numpy as np
> df.groupby(by=['A']).agg({'C':[np.mean, 'sum'], 'D':['count',np.std]})
      C     D
     mean sum count    std
A
a 108.000000 324   3 8.144528
b  95.000000 190   2 4.242641
c 104.666667 314   3 3.785939
 
 
>ps: 不同列使用多个不同函数进行聚合C： mean，sum；D：count，std

方法1：agg
> df.groupby(by=['A']).agg({'C':[np.mean]})
      C
     mean
A
a 108.000000
b  95.000000
c 104.666667
> type(df.groupby(by=['A']).agg({'C':[np.mean]}))
<class 'pandas.core.frame.DataFrame'>
 
 
方法2：索引
> df.groupby(by=['A'])['C'].mean()
A
a  108.000000
b   95.000000
c  104.666667
Name: C, dtype: float64
> type(df.groupby(by=['A'])['C'].mean())
<class 'pandas.core.series.Series'>
 
 
 
总结： 两种方法结果一样，但是一个类型是DataFrame,一个为Series；有时候会用上

> df
  A B  C  D
0 a 2 102  2
1 b 8  98 98
2 a 1 107 17
3 c 4 104 14
4 a 3 115 15
5 c 2  87  7
6 b 5  92 92
7 c 9 123 13
> df.sort_index(ascending=False)    ### 索引
  A B  C  D
7 c 9 123 13
6 b 5  92 92
5 c 2  87  7
4 a 3 115 15
3 c 4 104 14
2 a 1 107 17
1 b 8  98 98
0 a 2 102  2

> df.sort_values(by="A",ascending=False)    # 按某一列
  A B  C  D
3 c 4 104 14
5 c 2  87  7
7 c 9 123 13
1 b 8  98 98
6 b 5  92 92
0 a 2 102  2
2 a 1 107 17
4 a 3 115 15
 
> df.sort_values(by=["B","A"],ascending=False)  # 按2列
  A B  C  D
7 c 9 123 13
1 b 8  98 98
6 b 5  92 92
3 c 4 104 14
4 a 3 115 15
5 c 2  87  7
0 a 2 102  2
2 a 1 107 17