12-Groupby扩展操作--数据分析

groupby扩展操作

In [3]:

import pandas as pd

import numpy as np

df = pd.DataFrame({'A':['foo','bar','foo','bar','foo','bar','foo','foo'],

                  'B':['one','one','two','three','two','two','one','three'],

                  'C':np.random.randn(8),

                  'D':np.random.randn(8)})

df

Out[3]:

	A	B	C	D
0	foo	one	-0.363353	0.968036
1	bar	one	-0.713551	-0.540636
2	foo	two	-0.002899	-2.430552
3	bar	three	0.452300	-0.280095
4	foo	two	-1.083832	1.556935
5	bar	two	-0.266669	1.286793
6	foo	one	1.110641	-0.023415
7	foo	three	1.337744	0.463671

In [5]:

grouped = df.groupby('A')

grouped

Out[5]:

<pandas.core.groupby.groupby.DataFrameGroupBy object at 0x000000000530CEF0>

In [7]:

grouped.count()

Out[7]:

	B	C	D
A
bar	3	3	3
foo	5	5	5

In [8]:

grouped = df.groupby(['A','B','C'])

grouped.count()

Out[8]:

			D
A	B	C
bar	one	-0.713551	1
	three	0.452300	1
	two	-0.266669	1
foo	one	-0.363353	1
		1.110641	1
	three	1.337744	1
	two	-1.083832	1
		-0.002899	1

In [9]:

def get_letter_type(letter):

    if letter.lower() in 'aeiou':

        return 'a'

    else:

        return 'b'

grouped = df.groupby(get_letter_type,axis = 1)

grouped.count().iloc[0]

Out[9]:

a    1
b    3
Name: 0, dtype: int64

In [10]:

s = pd.Series([1,2,3,1,2,3],[8,7,5,8,7,5])

s

Out[10]:

8    1
7    2
5    3
8    1
7    2
5    3
dtype: int64

level=0 分组级别，其中0代表按照第一列分组。也可以直接写列的名字

In [13]:

grouped = s.groupby(level=0,sort=False)

In [14]:

grouped.first()

Out[14]:

8    1
7    2
5    3
dtype: int64

In [15]:

grouped.last()

Out[15]:

8    1
7    2
5    3
dtype: int64

In [16]:

grouped.sum()

Out[16]:

8    2
7    4
5    6
dtype: int64

In [17]:

df2 = pd.DataFrame({'X':['A','B','A','B'],

                   'Y':[1,2,3,4]})

df2

Out[17]:

	X	Y
0	A	1
1	B	2
2	A	3
3	B	4

In [19]:

df2.groupby(['X']).get_group('A')

Out[19]:

	X	Y
0	A	1
2	A	3

In [20]:

df2.groupby(['X']).get_group('B')

Out[20]:

	X	Y
1	B	2
3	B	4

In [21]:

arrays = [['bar', 'bar', 'baz', 'baz', 'foo', 'foo', 'qux', 'qux'],

          ['one', 'two', 'one', 'two', 'one', 'two', 'one', 'two']]

In [23]:

index = pd.MultiIndex.from_arrays(arrays,names=['first','second'])

In [24]:

s = pd.Series(np.random.randn(8),index=index)

s

Out[24]:

first  second
bar    one       1.032756
       two       1.155843
baz    one      -0.862770
       two       0.412540
foo    one       1.961790
       two       0.383244
qux    one       1.095939
       two       0.871055
dtype: float64

### groupby(level=0)  level属性分组级别，对应的值可以是数字，0代表第一列。或者直接写列的名字

In [25]:

grouped = s.groupby(level = 0)

grouped.sum()

Out[25]:

first
bar    2.188598
baz   -0.450230
foo    2.345034
qux    1.966994
dtype: float64

In [26]:

grouped = s.groupby(level = 1)

grouped.sum()

Out[26]:

second
one    3.227715
two    2.822682
dtype: float64

In [27]:

grouped = s.groupby(level = 'first')

grouped.sum()

Out[27]:

first
bar    2.188598
baz   -0.450230
foo    2.345034
qux    1.966994
dtype: float64

In [28]:

df

Out[28]:

	A	B	C	D
0	foo	one	-0.363353	0.968036
1	bar	one	-0.713551	-0.540636
2	foo	two	-0.002899	-2.430552
3	bar	three	0.452300	-0.280095
4	foo	two	-1.083832	1.556935
5	bar	two	-0.266669	1.286793
6	foo	one	1.110641	-0.023415
7	foo	three	1.337744	0.463671

In [29]:

grouped = df.groupby(['A','B'])

grouped.aggregate(np.sum)

Out[29]:

		C	D
A	B
bar	one	-0.713551	-0.540636
	three	0.452300	-0.280095
	two	-0.266669	1.286793
foo	one	0.747288	0.944621
	three	1.337744	0.463671
	two	-1.086731	-0.873618

In [30]:

grouped = df.groupby(['A','B'],as_index=False)

grouped.aggregate(np.sum)

Out[30]:

	A	B	C	D
0	bar	one	-0.713551	-0.540636
1	bar	three	0.452300	-0.280095
2	bar	two	-0.266669	1.286793
3	foo	one	0.747288	0.944621
4	foo	three	1.337744	0.463671
5	foo	two	-1.086731	-0.873618

分组求和后，进行重新构建索引

In [33]:

grouped = df.groupby(['A','B']).sum().reset_index()

grouped

Out[33]:

	A	B	C	D
0	bar	one	-0.713551	-0.540636
1	bar	three	0.452300	-0.280095
2	bar	two	-0.266669	1.286793
3	foo	one	0.747288	0.944621
4	foo	three	1.337744	0.463671
5	foo	two	-1.086731	-0.873618

In [34]:

grouped = df.groupby(['A','B'])

grouped.size()

Out[34]:

A    B    
bar  one      1
     three    1
     two      1
foo  one      2
     three    1
     two      2
dtype: int64

describe() 数据统计

In [35]:

grouped.describe().head()

Out[35]:

		C								D
		count	mean	std	min	25%	50%	75%	max	count	mean	std	min	25%	50%	75%	max
A	B
bar	one	1.0	-0.713551	NaN	-0.713551	-0.713551	-0.713551	-0.713551	-0.713551	1.0	-0.540636	NaN	-0.540636	-0.540636	-0.540636	-0.540636	-0.540636
	three	1.0	0.452300	NaN	0.452300	0.452300	0.452300	0.452300	0.452300	1.0	-0.280095	NaN	-0.280095	-0.280095	-0.280095	-0.280095	-0.280095
	two	1.0	-0.266669	NaN	-0.266669	-0.266669	-0.266669	-0.266669	-0.266669	1.0	1.286793	NaN	1.286793	1.286793	1.286793	1.286793	1.286793
foo	one	2.0	0.373644	1.042271	-0.363353	0.005145	0.373644	0.742142	1.110641	2.0	0.472310	0.701061	-0.023415	0.224448	0.472310	0.720173	0.968036
	three	1.0	1.337744	NaN	1.337744	1.337744	1.337744	1.337744	1.337744	1.0	0.463671	NaN	0.463671	0.463671	0.463671	0.463671	0.463671

agg([np.sum,np.mean,np.std]) 自定义统计参数

In [38]:

grouped = df.groupby('A')

grouped['C'].agg([np.sum,np.mean,np.std])

Out[38]:

	sum	mean	std
A
bar	-0.527920	-0.175973	0.588194
foo	0.998301	0.199660	1.016180

In [39]:

grouped['C'].agg({'求和':np.sum,'平均':np.mean,'方差':np.std})

d:\pythons\lib\site-packages\ipykernel_launcher.py:1: FutureWarning: using a dict on a Series for aggregation
is deprecated and will be removed in a future version
  """Entry point for launching an IPython kernel.

Out[39]:

	求和	平均	方差
A
bar	-0.527920	-0.175973	0.588194
foo	0.998301	0.199660	1.016180

In [ ]:

​