[雪峰磁针石博客]数据分析工具pandas快速入门教程2-pandas数据结构

  1. 云栖社区>
  2. 博客>
  3. 正文

[雪峰磁针石博客]数据分析工具pandas快速入门教程2-pandas数据结构

python人工智能命理 2018-08-23 13:15:48 浏览883
展开阅读全文

创建数据

Series和python的列表类似。DataFrame则类似值为Series的字典。

create.py


#!/usr/bin/env python3
# -*- coding: utf-8 -*-
# create.py

import pandas as pd

print("\n\n创建序列Series")
s = pd.Series(['banana', 42])
print(s)

print("\n\n指定索引index创建序列Series")
s = pd.Series(['Wes McKinney', 'Creator of Pandas'], index=['Person', 'Who'])
print(s)

# 注意:列名未必为执行的顺序,通常为按字母排序
print("\n\n创建数据帧DataFrame")
scientists = pd.DataFrame({
    ' Name': ['Rosaline Franklin', 'William Gosset'],
    ' Occupation': ['Chemist', 'Statistician'],
    ' Born': ['1920-07-25', '1876-06-13'],
    ' Died': ['1958-04-16', '1937-10-16'],
    ' Age': [37, 61]})
print(scientists)

print("\n\n指定顺序(index和columns)创建数据帧DataFrame")
scientists = pd.DataFrame(
    data={'Occupation': ['Chemist', 'Statistician'],
    'Born': ['1920-07-25', '1876-06-13'],
    'Died': ['1958-04-16', '1937-10-16'],
    'Age': [37, 61]},
    index=['Rosaline Franklin', 'William Gosset'],
    columns=['Occupation', 'Born', 'Died', 'Age'])
print(scientists)

执行结果:


$ ./create.py 


创建序列Series
0    banana
1        42
dtype: object


指定索引index创建序列Series
Person         Wes McKinney
Who       Creator of Pandas
dtype: object


创建数据帧DataFrame
                Name    Occupation        Born        Died   Age
0  Rosaline Franklin       Chemist  1920-07-25  1958-04-16    37
1     William Gosset  Statistician  1876-06-13  1937-10-16    61


指定顺序(index和columns)创建数据帧DataFrame
                     Occupation        Born        Died  Age
Rosaline Franklin       Chemist  1920-07-25  1958-04-16   37
William Gosset     Statistician  1876-06-13  1937-10-16   61

Series

官方文档:http://pandas.pydata.org/pandas-docs/stable/generated/pandas.Series.html

Series的属性

属性 描述
loc 使用索引值获取子集
iloc 使用索引位置获取子集
dtype或dtypes 类型
T 转置
shape 数据的尺寸
size 元素的数量
values ndarray或类似ndarray的Series

Series的方法

方法 描述
append 连接2个或更多系列
corr 计算与其他Series的关联
cov 与其他Series计算协方差
describe 计算汇总统计
drop duplicates 返回一个没有重复项的Series
equals Series是否具有相同的元素
get values 获取Series的值,与values属性相同
hist 绘制直方图
min 返回最小值
max 返回最大值
mean 返回算术平均值
median 返回中位数
mode(s) 返回mode(s)
replace 用指定值替换系列中的值
sample 返回Series中值的随机样本
sort values 排序
to frame 转换为数据帧
transpose 返回转置
unique 返回numpy.ndarray唯一值

series.py


#!/usr/bin/python3
# -*- coding: utf-8 -*-
# CreateDate: 2018-3-14
# series.py

import pandas as pd
import numpy as np

scientists = pd.DataFrame(
    data={'Occupation': ['Chemist', 'Statistician'],
    'Born': ['1920-07-25', '1876-06-13'],
    'Died': ['1958-04-16', '1937-10-16'],
    'Age': [37, 61]},
    index=['Rosaline Franklin', 'William Gosset'],
    columns=['Occupation', 'Born', 'Died', 'Age'])
print(scientists)

# 从数据帧(DataFrame)获取的行或者列为Series
first_row = scientists.loc['William Gosset']
print(type(first_row))
print(first_row)

# index和keys是一样的
print(first_row.index)
print(first_row.keys())
print(first_row.values)

print(first_row.index[0])
print(first_row.keys()[0])

# Pandas.Series和numpy.ndarray很类似
ages = scientists['Age']
print(ages)

# 统计,更多参考http://pandas.pydata.org/pandas-docs/stable/basics.html#descriptive-statistics
print(ages.mean())
print(ages.min())
print(ages.max())
print(ages.std())

scientists = pd.read_csv('../data/scientists.csv')
ages = scientists['Age']
print(ages)
print(ages.mean())
print(ages.describe())
print(ages[ages > ages.mean()])
print(ages > ages.mean())
manual_bool_values = [True, True, False, False, True, True, False, False]
print(ages[manual_bool_values])

print(ages + ages)
print(ages * ages)
print(ages + 100)
print(ages * 2)
print(ages + pd.Series([1, 100]))
# print(ages + np.array([1, 100])) 会报错,不同类型相加,大小一定要一样
print(ages + np.array([1, 100, 1, 100, 1, 100, 1, 100]))

# 排序: 默认有自动排序
print(ages)
rev_ages = ages.sort_index(ascending=False)
print(rev_ages)
print(ages * 2)
print(ages + rev_ages)

执行结果


$ python3 series.py 
                     Occupation        Born        Died  Age
Rosaline Franklin       Chemist  1920-07-25  1958-04-16   37
William Gosset     Statistician  1876-06-13  1937-10-16   61
<class 'pandas.core.series.Series'>
Occupation    Statistician
Born            1876-06-13
Died            1937-10-16
Age                     61
Name: William Gosset, dtype: object
Index(['Occupation', 'Born', 'Died', 'Age'], dtype='object')
Index(['Occupation', 'Born', 'Died', 'Age'], dtype='object')
['Statistician' '1876-06-13' '1937-10-16' 61]
Occupation
Occupation
Rosaline Franklin    37
William Gosset       61
Name: Age, dtype: int64
49.0
37
61
16.97056274847714
0    37
1    61
2    90
3    66
4    56
5    45
6    41
7    77
Name: Age, dtype: int64
59.125
count     8.000000
mean     59.125000
std      18.325918
min      37.000000
25%      44.000000
50%      58.500000
75%      68.750000
max      90.000000
Name: Age, dtype: float64
1    61
2    90
3    66
7    77
Name: Age, dtype: int64
0    False
1     True
2     True
3     True
4    False
5    False
6    False
7     True
Name: Age, dtype: bool
0    37
1    61
4    56
5    45
Name: Age, dtype: int64
0     74
1    122
2    180
3    132
4    112
5     90
6     82
7    154
Name: Age, dtype: int64
0    1369
1    3721
2    8100
3    4356
4    3136
5    2025
6    1681
7    5929
Name: Age, dtype: int64
0    137
1    161
2    190
3    166
4    156
5    145
6    141
7    177
Name: Age, dtype: int64
0     74
1    122
2    180
3    132
4    112
5     90
6     82
7    154
Name: Age, dtype: int64
0     38.0
1    161.0
2      NaN
3      NaN
4      NaN
5      NaN
6      NaN
7      NaN
dtype: float64
0     38
1    161
2     91
3    166
4     57
5    145
6     42
7    177
Name: Age, dtype: int64
0    37
1    61
2    90
3    66
4    56
5    45
6    41
7    77
Name: Age, dtype: int64
7    77
6    41
5    45
4    56
3    66
2    90
1    61
0    37
Name: Age, dtype: int64
0     74
1    122
2    180
3    132
4    112
5     90
6     82
7    154
Name: Age, dtype: int64
0     74
1    122
2    180
3    132
4    112
5     90
6     82
7    154
Name: Age, dtype: int64

数据帧(DataFrame)

DataFrame是最常见的Pandas对象,可认为是Python存储类似电子表格的数据的方式。Series多常见功能都包含在DataFrame中。

子集的方法

注意ix现在已经不推荐使用。

DataFrame常用的索引操作有:

方式 描述
df[val] 选择单个列
df [[ column1, column2, ... ]] 选择多个列
df.loc[val] 选择行
  1. loc [[ label1 , label2 ,...]] | 选择多行 |
    df.loc[:, val] | 基于行index选择列 |

df.loc[val1, val2] | 选择行列 |
df.iloc[row number] | 基于行数选择行 |

  1. iloc [[ row1, row2, ...]] Multiple rows by row number | 基于行数选择多行 |
    df.iloc[:, where] | 选择列 |

df.iloc[where_i, where_j] | 选择行列 |
df.at[label_i, label_j] | 选择值 |
df.iat[i, j] | 选择值 |
reindex method | 通过label选择多行或列 |
get_value, set_value | 通过label选择耽搁行或列
df[bool] | 选择行
df [[ bool1, bool2, ...]] | 选择行
df[ start :stop: step ] | 基于行数选择行

#!/usr/bin/python3
# -*- coding: utf-8 -*-
# CreateDate: 2018-3-31
# df.py

import pandas as pd
import numpy as np

scientists = pd.read_csv('../data/scientists.csv')
print(scientists[scientists['Age'] > scientists['Age'].mean()])
first_half = scientists[: 4]
second_half = scientists[ 4 :]
print(first_half)
print(second_half)
print(first_half + second_half)
print(scientists * 2)

执行结果


#!/usr/bin/python3
# -*- coding: utf-8 -*-
# df.py

import pandas as pd
import numpy as np

scientists = pd.read_csv('../data/scientists.csv')
print(scientists[scientists['Age'] > scientists['Age'].mean()])
first_half = scientists[: 4]
second_half = scientists[ 4 :]
print(first_half)
print(second_half)
print(first_half + second_half)
print(scientists * 2)

执行结果


$ python3 df.py 
                   Name        Born        Died  Age     Occupation
1        William Gosset  1876-06-13  1937-10-16   61   Statistician
2  Florence Nightingale  1820-05-12  1910-08-13   90          Nurse
3           Marie Curie  1867-11-07  1934-07-04   66        Chemist
7          Johann Gauss  1777-04-30  1855-02-23   77  Mathematician
                   Name        Born        Died  Age    Occupation
0     Rosaline Franklin  1920-07-25  1958-04-16   37       Chemist
1        William Gosset  1876-06-13  1937-10-16   61  Statistician
2  Florence Nightingale  1820-05-12  1910-08-13   90         Nurse
3           Marie Curie  1867-11-07  1934-07-04   66       Chemist
            Name        Born        Died  Age          Occupation
4  Rachel Carson  1907-05-27  1964-04-14   56           Biologist
5      John Snow  1813-03-15  1858-06-16   45           Physician
6    Alan Turing  1912-06-23  1954-06-07   41  Computer Scientist
7   Johann Gauss  1777-04-30  1855-02-23   77       Mathematician
  Name Born Died  Age Occupation
0  NaN  NaN  NaN  NaN        NaN
1  NaN  NaN  NaN  NaN        NaN
2  NaN  NaN  NaN  NaN        NaN
3  NaN  NaN  NaN  NaN        NaN
4  NaN  NaN  NaN  NaN        NaN
5  NaN  NaN  NaN  NaN        NaN
6  NaN  NaN  NaN  NaN        NaN
7  NaN  NaN  NaN  NaN        NaN
                                       Name                  Born  \
0        Rosaline FranklinRosaline Franklin  1920-07-251920-07-25   
1              William GossetWilliam Gosset  1876-06-131876-06-13   
2  Florence NightingaleFlorence Nightingale  1820-05-121820-05-12   
3                    Marie CurieMarie Curie  1867-11-071867-11-07   
4                Rachel CarsonRachel Carson  1907-05-271907-05-27   
5                        John SnowJohn Snow  1813-03-151813-03-15   
6                    Alan TuringAlan Turing  1912-06-231912-06-23   
7                  Johann GaussJohann Gauss  1777-04-301777-04-30   

                   Died  Age                            Occupation  
0  1958-04-161958-04-16   74                        ChemistChemist  
1  1937-10-161937-10-16  122              StatisticianStatistician  
2  1910-08-131910-08-13  180                            NurseNurse  
3  1934-07-041934-07-04  132                        ChemistChemist  
4  1964-04-141964-04-14  112                    BiologistBiologist  
5  1858-06-161858-06-16   90                    PhysicianPhysician  
6  1954-06-071954-06-07   82  Computer ScientistComputer Scientist  
7  1855-02-231855-02-23  154            MathematicianMathematician  

修改列

#!/usr/bin/python3
# -*- coding: utf-8 -*-
# Author:    xurongzhong#126.com wechat:pythontesting qq:37391319
# qq群:144081101 591302926  567351477
# CreateDate: 2018-06-07
# change.py

import pandas as pd
import numpy as np
import random

scientists = pd.read_csv('../data/scientists.csv')
print(scientists['Born'].dtype)
print(scientists['Died'].dtype)
print(scientists.head())

# 转为日期 参考:https://docs.python.org/3.5/library/datetime.html
born_datetime = pd.to_datetime(scientists['Born'], format='%Y-%m-%d')
died_datetime = pd.to_datetime(scientists['Died'], format='%Y-%m-%d')
# 增加列
scientists['born_dt'], scientists['died_dt'] = (born_datetime, died_datetime)
print(scientists.shape)
print(scientists.head())

random.seed(42)
random.shuffle(scientists['Age']) # 此修改会作用于scientists
print(scientists.head())

scientists['age_days_dt'] = (scientists['died_dt'] - scientists['born_dt'])
print(scientists.head())

执行结果:

$ python3 change.py 
object
object
                   Name        Born        Died  Age    Occupation
0     Rosaline Franklin  1920-07-25  1958-04-16   37       Chemist
1        William Gosset  1876-06-13  1937-10-16   61  Statistician
2  Florence Nightingale  1820-05-12  1910-08-13   90         Nurse
3           Marie Curie  1867-11-07  1934-07-04   66       Chemist
4         Rachel Carson  1907-05-27  1964-04-14   56     Biologist
(8, 7)
                   Name        Born        Died  Age    Occupation    born_dt  \
0     Rosaline Franklin  1920-07-25  1958-04-16   37       Chemist 1920-07-25   
1        William Gosset  1876-06-13  1937-10-16   61  Statistician 1876-06-13   
2  Florence Nightingale  1820-05-12  1910-08-13   90         Nurse 1820-05-12   
3           Marie Curie  1867-11-07  1934-07-04   66       Chemist 1867-11-07   
4         Rachel Carson  1907-05-27  1964-04-14   56     Biologist 1907-05-27   

     died_dt  
0 1958-04-16  
1 1937-10-16  
2 1910-08-13  
3 1934-07-04  
4 1964-04-14  
/usr/lib/python3.5/random.py:272: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame

See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy
  x[i], x[j] = x[j], x[i]
                   Name        Born        Died  Age    Occupation    born_dt  \
0     Rosaline Franklin  1920-07-25  1958-04-16   66       Chemist 1920-07-25   
1        William Gosset  1876-06-13  1937-10-16   56  Statistician 1876-06-13   
2  Florence Nightingale  1820-05-12  1910-08-13   41         Nurse 1820-05-12   
3           Marie Curie  1867-11-07  1934-07-04   77       Chemist 1867-11-07   
4         Rachel Carson  1907-05-27  1964-04-14   90     Biologist 1907-05-27   

     died_dt  
0 1958-04-16  
1 1937-10-16  
2 1910-08-13  
3 1934-07-04  
4 1964-04-14  
                   Name        Born        Died  Age    Occupation    born_dt  \
0     Rosaline Franklin  1920-07-25  1958-04-16   66       Chemist 1920-07-25   
1        William Gosset  1876-06-13  1937-10-16   56  Statistician 1876-06-13   
2  Florence Nightingale  1820-05-12  1910-08-13   41         Nurse 1820-05-12   
3           Marie Curie  1867-11-07  1934-07-04   77       Chemist 1867-11-07   
4         Rachel Carson  1907-05-27  1964-04-14   90     Biologist 1907-05-27   

     died_dt age_days_dt  
0 1958-04-16  13779 days  
1 1937-10-16  22404 days  
2 1910-08-13  32964 days  
3 1934-07-04  24345 days  
4 1964-04-14  20777 days  

数据导入导出

out.py

#!/usr/bin/python3
# -*- coding: utf-8 -*-
# Author:    china-testing#126.com wechat:pythontesting qq群:630011153
# CreateDate: 2018-3-31
# out.py

import pandas as pd
import numpy as np
import random
scientists = pd.read_csv('../data/scientists.csv')

names = scientists['Name']
print(names)
names.to_pickle('../output/scientists_names_series.pickle')
scientists.to_pickle('../output/scientists_df.pickle')

# .p, .pkl,  .pickle 是常用的pickle文件扩展名
scientist_names_from_pickle = pd.read_pickle('../output/scientists_df.pickle')
print(scientist_names_from_pickle)

names.to_csv('../output/scientist_names_series.csv')
scientists.to_csv('../output/scientists_df.tsv', sep='\t')
# 不输出行号
scientists.to_csv('../output/scientists_df_no_index.csv', index=None)

# Series可以转为df再输出成excel文件
names_df = names.to_frame()
names_df.to_excel('../output/scientists_names_series_df.xls')
names_df.to_excel('../output/scientists_names_series_df.xlsx')

scientists.to_excel('../output/scientists_df.xlsx', sheet_name='scientists',
                    index=False)
                    

执行结果:

$ python3 out.py 
0       Rosaline Franklin
1          William Gosset
2    Florence Nightingale
3             Marie Curie
4           Rachel Carson
5               John Snow
6             Alan Turing
7            Johann Gauss
Name: Name, dtype: object
                   Name        Born        Died  Age          Occupation
0     Rosaline Franklin  1920-07-25  1958-04-16   37             Chemist
1        William Gosset  1876-06-13  1937-10-16   61        Statistician
2  Florence Nightingale  1820-05-12  1910-08-13   90               Nurse
3           Marie Curie  1867-11-07  1934-07-04   66             Chemist
4         Rachel Carson  1907-05-27  1964-04-14   56           Biologist
5             John Snow  1813-03-15  1858-06-16   45           Physician
6           Alan Turing  1912-06-23  1954-06-07   41  Computer Scientist
7          Johann Gauss  1777-04-30  1855-02-23   77       Mathematician
    

注意:序列一般是直接输出成excel文件

更多的输入输出方法:

方式 描述
to_clipboard 将数据保存到系统剪贴板进行粘贴
to_dense 将数据转换为常规“密集”DataFrame
to_dict 将数据转换为Python字典
to_gbq 将数据转换为Google BigQuery表格
toJidf 将数据保存为分层数据格式(HDF)
to_msgpack 将数据保存到可移植的类似JSON的二进制文件中
toJitml 将数据转换为HTML表格
tojson 将数据转换为JSON字符串
toJatex 将数据转换为LTEXtabular环境
to_records 将数据转换为记录数组
to_string 将DataFrame显示为stdout的字符串
to_sparse 将数据转换为SparceDataFrame
to_sql 将数据保存到SQL数据库中
to_stata 将数据转换为Stata dta文件
  • 读CSV文件

read_csv.py


#!/usr/bin/python3
# -*- coding: utf-8 -*-
# Author:    china-testing#126.com wechat:pythontesting QQ群:630011153
# CreateDate: 2018-3-9
# read_csv.py

import pandas as pd

df = pd.read_csv("1.csv", header=None) # 不读取列名
print("df:")
print(df)

print("df.head():")
print(df.head()) # head(self, n=5),默认为5行,类似的有tail
print("df.tail():")
print(df.tail())

df = pd.read_csv("1.csv") # 默认读取列名
print("df:")
print(df)

df = pd.read_csv("1.csv", names=['号码','群号']) # 自定义列名
print("df:")
print(df)

# 自定义列名,去掉第一行
df = pd.read_csv("1.csv", skiprows=[0], names=['号码','群号'])
print("df:")
print(df)

执行结果:


df:
           0          1
0         qq    qqgroup
1   37391319  144081101
2   37391320  144081102
3   37391321  144081103
4   37391322  144081104
5   37391323  144081105
6   37391324  144081106
7   37391325  144081107
8   37391326  144081108
9   37391327  144081109
10  37391328  144081110
11  37391329  144081111
12  37391330  144081112
13  37391331  144081113
14  37391332  144081114
15  37391333  144081115
df.head():
          0          1
0        qq    qqgroup
1  37391319  144081101
2  37391320  144081102
3  37391321  144081103
4  37391322  144081104
df.tail():
           0          1
11  37391329  144081111
12  37391330  144081112
13  37391331  144081113
14  37391332  144081114
15  37391333  144081115
df:
          qq    qqgroup
0   37391319  144081101
1   37391320  144081102
2   37391321  144081103
3   37391322  144081104
4   37391323  144081105
5   37391324  144081106
6   37391325  144081107
7   37391326  144081108
8   37391327  144081109
9   37391328  144081110
10  37391329  144081111
11  37391330  144081112
12  37391331  144081113
13  37391332  144081114
14  37391333  144081115
df:
          号码         群号
0         qq    qqgroup
1   37391319  144081101
2   37391320  144081102
3   37391321  144081103
4   37391322  144081104
5   37391323  144081105
6   37391324  144081106
7   37391325  144081107
8   37391326  144081108
9   37391327  144081109
10  37391328  144081110
11  37391329  144081111
12  37391330  144081112
13  37391331  144081113
14  37391332  144081114
15  37391333  144081115
df:
          号码         群号
0   37391319  144081101
1   37391320  144081102
2   37391321  144081103
3   37391322  144081104
4   37391323  144081105
5   37391324  144081106
6   37391325  144081107
7   37391326  144081108
8   37391327  144081109
9   37391328  144081110
10  37391329  144081111
11  37391330  144081112
12  37391331  144081113
13  37391332  144081114
14  37391333  144081115
  • 写CSV文件
#!/usr/bin/python3
# -*- coding: utf-8 -*-
# write_csv.py

import pandas as pd

data ={'qq': [37391319,37391320], 'group':[1,2]}

df = pd.DataFrame(data=data, columns=['qq','group'])
df.to_csv('2.csv',index=False)

读写excel和csv类似,不过要改用read_excel来读,excel_summary_demo, 提供了多个excel求和的功能,可以做为excel读写的实例,这里不再赘述。

参考资料

网友评论

登录后评论
0/500
评论
python人工智能命理
+ 关注