数据类型-DataFrame

• DataFrame是由多个Series数据列组成的表格数据类型，每行Series值都增加了一个共用的索引
• 既有行索引，又有列索引
  
  • 行索引，表明不同行，横向索引，叫index，0轴，axis=0
  • 列索引，表名不同列，纵向索引，叫columns，1轴，axis=1
• DataFrame数据类型可视为：二维 带标签 数组
• 每列值的类型可以不同
• 基本操作类似Series，依据行列索引操作
• 常用于表达二维数据，但也可以表达多维数据(Dataframe嵌套，极少用)

DataFrame数据类型创建

Python list列表 创建DataFrame

import pandas as pd

df = pd.DataFrame([True, 1, 2.3, 'a', '你好']) # 1维
df

df = pd.DataFrame([[True,1,2.3,'a','你好'],[1,2,3,4,5]]) #2维
df

# 3维，不建议
df = pd.DataFrame([[[True,1,2.3,'a','你好'],
                    [1,2,3,4,5]],
                   [[True,1,2.3,'a','你好'],
                    [1,2,3,4,5]]
                  ]) 
df

Python 字典 创建DataFrame

df = pd.DataFrame({'one':[1,2,3,4],
                   'two':[9,8,7,6]})
df

# 自定义行索引
df = pd.DataFrame({'one':[1,2,3,4],
                   'two':[9,8,7,6]},index = ['a','b','c','d']) 
df

df = pd.DataFrame({
    'A' : 1,
    'B' : 2.3,
    'C' : ['x','y',5] #需要多行
})
df

dt = {
    'one' : pd.Series([1,2,3],index=['a','b','c']),
    'two' : pd.Series([9,8,7,6],index=['a','b','c','d',])
}
dt

{'one': a    1
 b    2
 c    3
 dtype: int64, 'two': a    9
 b    8
 c    7
 d    6
 dtype: int64}

# one two自动列索引,abcd自动行索引.每个元素对应DataFrame的一列,每个元素内的键值对应一行
d = pd.DataFrame(dt) 
d

# 数据根据行列索引自动补齐
d_2 = pd.DataFrame(dt,index=['b','c','d'],columns=['two','three']) 
d_2

ndarray数组 创建DataFrame

import numpy as np

df = pd.DataFrame(np.arange(10).reshape(2,5)) # 自动生成行/列索引
df

# 自定义行列索引
df = pd.DataFrame(np.random.randn(6,4),
                  index=[1,2,3,4,5,6],
                  columns=['a','b','c','d']) 
df

Series 创建DataFrame

e = pd.DataFrame([pd.Series([1,2,3]),
                  pd.Series([9,8,7,6])],
                 index=['a','b'])
e

DataFrame属性

di = {
    '姓名':['张三','李四','王五','赵六'],
    '性别':['男','女','女','男'],
    '年龄':[12,22,32,42],
    '地址':['北京','上海','广州','深圳']
}
di

{'地址': ['北京', '上海', '广州', '深圳'],
 '姓名': ['张三', '李四', '王五', '赵六'],
 '年龄': [12, 22, 32, 42],
 '性别': ['男', '女', '女', '男']}

d = pd.DataFrame(di,index=['d1','d2','d3','d4'])
d

d.head() # 显示头部几行

d.tail(3) # 显示末尾几行

d.info() # 相关信息概览

<class 'pandas.core.frame.DataFrame'>
Index: 4 entries, d1 to d4
Data columns (total 4 columns):
地址    4 non-null object
姓名    4 non-null object
年龄    4 non-null int64
性别    4 non-null object
dtypes: int64(1), object(3)
memory usage: 160.0+ bytes

d.shape # 行数 列数

(4, 4)

d.dtypes # 列数据类型

地址    object
姓名    object
年龄     int64
性别    object
dtype: object

d.index # 获取行索引

Index(['d1', 'd2', 'd3', 'd4'], dtype='object')

d.columns # 获取列索引

Index(['地址', '姓名', '年龄', '性别'], dtype='object')

d.values # 获取值

array([['北京', '张三', 12, '男'],
       ['上海', '李四', 22, '女'],
       ['广州', '王五', 32, '女'],
       ['深圳', '赵六', 42, '男']], dtype=object)

DataFrame查增改删

查 Read

类list/ndarray数据访问方式

dates = pd.date_range('20130101',periods=10)
dates

DatetimeIndex(['2013-01-01', '2013-01-02', '2013-01-03', '2013-01-04',
               '2013-01-05', '2013-01-06', '2013-01-07', '2013-01-08',
               '2013-01-09', '2013-01-10'],
              dtype='datetime64[ns]', freq='D')

df = pd.DataFrame(np.random.randn(10,4),index=dates,columns=['A','B','C','D'])
df

#索引
df['A']

2013-01-01    0.754077
2013-01-02    0.103394
2013-01-03    0.174730
2013-01-04   -0.950517
2013-01-05    0.076178
2013-01-06    1.371702
2013-01-07    0.126720
2013-01-08   -1.246918
2013-01-09    0.941099
2013-01-10    1.951555
Freq: D, Name: A, dtype: float64

df.A

2013-01-01    0.754077
2013-01-02    0.103394
2013-01-03    0.174730
2013-01-04   -0.950517
2013-01-05    0.076178
2013-01-06    1.371702
2013-01-07    0.126720
2013-01-08   -1.246918
2013-01-09    0.941099
2013-01-10    1.951555
Freq: D, Name: A, dtype: float64

df['A']['2013-01-01'] # 先列后行

0.75407705661157032

df.A['2013-01-01']

0.75407705661157032

df[['A','C']]

Pandas专用的数据访问方式 — .loc 通过自定义索引获取数据

#选取某行
df.loc['2013-01-01']

A    0.754077
B   -0.346202
C   -0.557050
D    0.778106
Name: 2013-01-01 00:00:00, dtype: float64

#选取某列
df.loc[:,'A']

2013-01-01    0.754077
2013-01-02    0.103394
2013-01-03    0.174730
2013-01-04   -0.950517
2013-01-05    0.076178
2013-01-06    1.371702
2013-01-07    0.126720
2013-01-08   -1.246918
2013-01-09    0.941099
2013-01-10    1.951555
Freq: D, Name: A, dtype: float64

# 选取特定值
df.loc['2013-01-01','A'] # 先行后列

0.75407705661157032

# 选取指定的行/列
df.loc[[dates[0],dates[2]],:] # 指定行

df.loc[:,['A','B']] # 指定列

df.loc[[dates[0],dates[2]],['A','B']] # 指定行列

# 切片
df.loc['2013-01-01':'2013-01-04',:] # 对行切片

df.loc[:,'A':'C'] # 对列切片

# 切片选取连续区块。行，列。左开右闭
df.loc['2013-01-01':'2013-01-04','A':'C'] 

.iloc 通过默认索引获取数据

# 选取某行
df.iloc[3]

A   -0.950517
B   -0.226887
C   -0.097138
D   -0.442010
Name: 2013-01-04 00:00:00, dtype: float64

# 选取某列
df.iloc[:,2]

2013-01-01   -0.557050
2013-01-02   -0.413054
2013-01-03    1.781379
2013-01-04   -0.097138
2013-01-05    1.142290
2013-01-06   -1.470106
2013-01-07   -2.212507
2013-01-08    1.761499
2013-01-09    1.927863
2013-01-10   -0.171690
Freq: D, Name: C, dtype: float64

# 选取特定值:
df.iloc[1,2]

-0.41305425875508139

# 选取指定的行/列
df.iloc[[1,2,4],:] # 指定行

df.iloc[:,[0,2]] # 指定列

df.iloc[[1,2,4],[0,2]] # 指定行列 ，先行后列

# 切片
df.iloc[1:3,:] # 对行切片:

df.iloc[:,1:3] # 对列切片:

df.iloc[3:5,0:2] # 切片选取连续区块。行，列。左开右闭

Boolean索引

# 通过某列选择数据:
df[df.A > 0]

# 通过where选择数据:
b = df[df > 0]
b

type(b['A']['2013-01-01'])

numpy.float64

# 通过 isin() 过滤数据:
df2 = df.copy()
df2['E'] = ['one', 'one','two','three','four','three','five','four','three','five']
df2

df2['E'].isin(['one','four'])

2013-01-01     True
2013-01-02     True
2013-01-03    False
2013-01-04    False
2013-01-05     True
2013-01-06    False
2013-01-07    False
2013-01-08     True
2013-01-09    False
2013-01-10    False
Freq: D, Name: E, dtype: bool

df2[df2['E'].isin(['one','four'])]

增 Create

s1 = pd.Series([1,2,3,4,5,6], 
               index=pd.date_range('20130102', periods=6))
s1

2013-01-02    1
2013-01-03    2
2013-01-04    3
2013-01-05    4
2013-01-06    5
2013-01-07    6
Freq: D, dtype: int64

# 新增一列数据
df2['F'] = s1
df2

改 Update

# 更新一列值
df2.loc[:,'D']

2013-01-01    0.778106
2013-01-02    0.268955
2013-01-03    1.643397
2013-01-04   -0.442010
2013-01-05   -0.952401
2013-01-06   -0.113098
2013-01-07    1.050036
2013-01-08    0.940741
2013-01-09   -0.549143
2013-01-10    0.869293
Freq: D, Name: D, dtype: float64

df2.loc[:,'D'] = 5
df2

df2.iloc[1,3]

5

df2.iloc[1,3] = 10.1
df2

# 通过where更新
df3 = df.copy()
df3[df3 > 0] = -df3
df3