Pandas高阶--第一节 层级索引、分组与聚合介绍、GroupBy对象及常用的聚合操作、自定义分组及聚合操作

 

 

 

 

实验：

 

第6课 数据分析工具

第一节 层级索引

In [30]:

 

import pandas as pd

import numpy as np

In [2]:

 

# 文件路径

filepath = r'C:\Users\ML Learning\Projects\第四章-数据分析预习内容\第四章-数据分析预习内容\第二节-数据分析工具pandas高阶\3_lesson_06\lesson_06\examples\datasets\2016_happiness.csv'

读取文件

In [3]:

 

data = pd.read_csv(filepath,usecols=['Country','Region','Happiness Rank','Happiness Score'])

In [7]:

 

# 数据预览

data.head()

Out[7]:

		Happiness Rank	Happiness Score
Region	Country
Western Europe	Denmark	1	7.526
	Switzerland	2	7.509
	Iceland	3	7.501
	Norway	4	7.498
	Finland	5	7.413

设置多个索引列

In [5]:

 

data.set_index(['Region','Country'],inplace=True)

In [9]:

 

data

Out[9]:

		Happiness Rank	Happiness Score
Region	Country
Western Europe	Denmark	1	7.526
	Switzerland	2	7.509
	Iceland	3	7.501
	Norway	4	7.498
	Finland	5	7.413
...	...	...	...
Sub-Saharan Africa	Benin	153	3.484
Southern Asia	Afghanistan	154	3.360
Sub-Saharan Africa	Togo	155	3.303
Middle East and Northern Africa	Syria	156	3.069
Sub-Saharan Africa	Burundi	157	2.905

157 rows × 2 columns

选取子集

In [11]:

 

# 外层选取

data.loc['Australia and New Zealand','New Zealand']

Out[11]:

Happiness Rank     8.000
Happiness Score    7.334
Name: (Australia and New Zealand, New Zealand), dtype: float64

交换层级顺序

In [12]:

 

data.swaplevel()

Out[12]:

		Happiness Rank	Happiness Score
Country	Region
Denmark	Western Europe	1	7.526
Switzerland	Western Europe	2	7.509
Iceland	Western Europe	3	7.501
Norway	Western Europe	4	7.498
Finland	Western Europe	5	7.413
...	...	...	...
Benin	Sub-Saharan Africa	153	3.484
Afghanistan	Southern Asia	154	3.360
Togo	Sub-Saharan Africa	155	3.303
Syria	Middle East and Northern Africa	156	3.069
Burundi	Sub-Saharan Africa	157	2.905

157 rows × 2 columns

层级索引排序

In [14]:

 

data.sort_index()

Out[14]:

		Happiness Rank	Happiness Score
Region	Country
Australia and New Zealand	Australia	9	7.313
	New Zealand	8	7.334
Central and Eastern Europe	Albania	109	4.655
	Armenia	121	4.360
	Azerbaijan	81	5.291
...	...	...	...
Western Europe	Portugal	94	5.123
	Spain	37	6.361
	Sweden	10	7.291
	Switzerland	2	7.509
	United Kingdom	23	6.725

157 rows × 2 columns

groupby()

In [15]:

 

# 按单列分组

obj1 = data.groupby('Region')

print(type(obj1))

<class 'pandas.core.groupby.generic.DataFrameGroupBy'>

In [16]:

 

#按多列分组

obj2 = data.groupby(['Region','Country'])

print(type(obj2))

<class 'pandas.core.groupby.generic.DataFrameGroupBy'>

常用的聚合操作

In [17]:

 

obj1.mean()

Out[17]:

	Happiness Rank	Happiness Score
Region
Australia and New Zealand	8.500000	7.323500
Central and Eastern Europe	78.448276	5.370690
Eastern Asia	67.166667	5.624167
Latin America and Caribbean	48.333333	6.101750
Middle East and Northern Africa	78.105263	5.386053
North America	9.500000	7.254000
Southeastern Asia	80.000000	5.338889
Southern Asia	111.714286	4.563286
Sub-Saharan Africa	129.657895	4.136421
Western Europe	29.190476	6.685667

In [18]:

 

type(obj1.mean())

Out[18]:

pandas.core.frame.DataFrame

In [19]:

 

obj1.max()

Out[19]:

	Happiness Rank	Happiness Score
Region
Australia and New Zealand	9	7.334
Central and Eastern Europe	129	6.596
Eastern Asia	101	6.379
Latin America and Caribbean	136	7.087
Middle East and Northern Africa	156	7.267
North America	13	7.404
Southeastern Asia	140	6.739
Southern Asia	154	5.196
Sub-Saharan Africa	157	5.648
Western Europe	99	7.526

In [20]:

 

obj1.size()

Out[20]:

Region
Australia and New Zealand           2
Central and Eastern Europe         29
Eastern Asia                        6
Latin America and Caribbean        24
Middle East and Northern Africa    19
North America                       2
Southeastern Asia                   9
Southern Asia                       7
Sub-Saharan Africa                 38
Western Europe                     21
dtype: int64

In [23]:

 

obj1.count()  # 按列统计，获取非空值

Out[23]:

	Happiness Rank	Happiness Score
Region
Australia and New Zealand	2	2
Central and Eastern Europe	29	29
Eastern Asia	6	6
Latin America and Caribbean	24	24
Middle East and Northern Africa	19	19
North America	2	2
Southeastern Asia	9	9
Southern Asia	7	7
Sub-Saharan Africa	38	38
Western Europe	21	21

In [24]:

 

obj2.size()

Out[24]:

Region                      Country       
Australia and New Zealand   Australia         1
                            New Zealand       1
Central and Eastern Europe  Albania           1
                            Armenia           1
                            Azerbaijan        1
                                             ..
Western Europe              Portugal          1
                            Spain             1
                            Sweden            1
                            Switzerland       1
                            United Kingdom    1
Length: 157, dtype: int64

自定义分组

In [25]:

 

# 自定义分组

def get_score_group(score):

    if score <= 4:

        score_group = 'low'

    elif score <= 6:

        score_group = 'middle'

    else:

        score_group = 'high'

    return score_group

In [26]:

 

# 方法1：传入自定义的函数进行分组按单列分组

data2 = data.set_index('Happiness Score')

data2.groupby(get_score_group).size()

Out[26]:

high      47
low       21
middle    89
dtype: int64

In [27]:

 

# 方法2：人为构造出一个分组列

data['score group'] = data['Happiness Score'].apply(get_score_group)

data.head()

Out[27]:

		Happiness Rank	Happiness Score	score group
Region	Country
Western Europe	Denmark	1	7.526	high
	Switzerland	2	7.509	high
	Iceland	3	7.501	high
	Norway	4	7.498	high
	Finland	5	7.413	high

In [28]:

 

data.groupby('Region').max()

Out[28]:

	Happiness Rank	Happiness Score	score group
Region
Australia and New Zealand	9	7.334	high
Central and Eastern Europe	129	6.596	middle
Eastern Asia	101	6.379	middle
Latin America and Caribbean	136	7.087	middle
Middle East and Northern Africa	156	7.267	middle
North America	13	7.404	high
Southeastern Asia	140	6.739	middle
Southern Asia	154	5.196	middle
Sub-Saharan Africa	157	5.648	middle
Western Europe	99	7.526	middle

In [31]:

 

data.groupby('Region').agg(np.max)

Out[31]:

	Happiness Rank	Happiness Score	score group
Region
Australia and New Zealand	9	7.334	high
Central and Eastern Europe	129	6.596	middle
Eastern Asia	101	6.379	middle
Latin America and Caribbean	136	7.087	middle
Middle East and Northern Africa	156	7.267	middle
North America	13	7.404	high
Southeastern Asia	140	6.739	middle
Southern Asia	154	5.196	middle
Sub-Saharan Africa	157	5.648	middle
Western Europe	99	7.526	middle

In [33]:

 

# 传入包含多个函数的列表

data.groupby('Region')['Happiness Score'].agg([np.max,np.min,np.mean])

Out[33]:

	amax	amin	mean
Region
Australia and New Zealand	7.334	7.313	7.323500
Central and Eastern Europe	6.596	4.217	5.370690
Eastern Asia	6.379	4.907	5.624167
Latin America and Caribbean	7.087	4.028	6.101750
Middle East and Northern Africa	7.267	3.069	5.386053
North America	7.404	7.104	7.254000
Southeastern Asia	6.739	3.907	5.338889
Southern Asia	5.196	3.360	4.563286
Sub-Saharan Africa	5.648	2.905	4.136421
Western Europe	7.526	5.033	6.685667

In [34]:

 

# 通过字典为每个列指定不同的操作方法

data.groupby('Region').agg({'Happiness Score':np.mean,'Happiness Rank': np.max})

Out[34]:

	Happiness Score	Happiness Rank
Region
Australia and New Zealand	7.323500	9
Central and Eastern Europe	5.370690	129
Eastern Asia	5.624167	101
Latin America and Caribbean	6.101750	136
Middle East and Northern Africa	5.386053	156
North America	7.254000	13
Southeastern Asia	5.338889	140
Southern Asia	4.563286	154
Sub-Saharan Africa	4.136421	157
Western Europe	6.685667	99

In [35]:

 

# 传入自定义函数

def max_min_diff(x):

    return x.max() - x.min()

​

data.groupby('Region')['Happiness Rank'].agg(max_min_diff)

Out[35]:

Region
Australia and New Zealand            1
Central and Eastern Europe         102
Eastern Asia                        67
Latin America and Caribbean        122
Middle East and Northern Africa    145
North America                        7
Southeastern Asia                  118
Southern Asia                       70
Sub-Saharan Africa                  91
Western Europe                      98
Name: Happiness Rank, dtype: int64

In [ ]:

 

​