Pandas的groupby的功能很强大，下面介绍groupby的用法：

首先，引入相关package：

import pandas as pd
import numpy as np

groupby的基础操作

按类别列分组（groupby），获取其他列的和

按多列进行分组（groupby）

聚合方法size()和count()

size跟count的区别： size计数时包含NaN值，而count不包含NaN值

In [10]: df = pd.DataFrame({"Name":["Alice", "Bob", "Mallory", "Mallory", "Bob" , "Mallory"],
    ...:                  "City":["Seattle", "Seattle", "Portland", "Seattle", "Seattle", "Portland"],
    ...:                  "Val":[4,3,3,np.nan,np.nan,4]})
    ...: 
    ...: df
    ...: 
Out[10]: 
       City     Name  Val
0   Seattle    Alice  4.0
1   Seattle      Bob  3.0
2  Portland  Mallory  3.0
3   Seattle  Mallory  NaN
4   Seattle      Bob  NaN
5  Portland  Mallory  4.0

count()

In [11]: df.groupby(["Name", "City"], as_index=False)['Val'].count()

Out[11]: 
      Name      City  Val
0    Alice   Seattle    1
1      Bob   Seattle    1
2  Mallory  Portland    2
3  Mallory   Seattle    0

size()

In [12]: df.groupby(["Name", "City"])['Val'].size().reset_index(name='Size')

Out[12]: 
      Name      City  Size
0    Alice   Seattle     1
1      Bob   Seattle     2
2  Mallory  Portland     2
3  Mallory   Seattle     1