数据挖掘---- pandas之索引

26 篇文章 3 订阅
7 篇文章 0 订阅


索引
Series的索引(obj[…])与Numpy数组索引的功能类似,只不过Series的索引值可以不是整数。

import numpy as np
import pandas as pd
obj=pd.Series(np.arange(4.),index=['a','b','c','d'])
obj

普通的python切片不包含尾部

a    0.0
b    1.0
c    2.0
d    3.0
dtype: float64

使用这些方法会修改Series的相应部分:

obj['b':'c']=5
obj
a    0.0
b    5.0
c    5.0
d    3.0
dtype: float64
import numpy as np
import pandas as pd
df = pd.read_csv('data/table.csv',index_col='ID')
df.head()
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1101S_1C_1Mstreet_11736334.0A+
1102S_1C_1Fstreet_21927332.5B+
1103S_1C_1Mstreet_21868287.2B+
1104S_1C_1Fstreet_21678180.4B-
1105S_1C_1Fstreet_41596484.8B+

使用单个值或者序列,可以从DataFrame中索引出一个或多个列

df['School']
df[['School','Class']]

这种索引也有特殊案例。首先可以根据一个布尔数组切片或者选择数据:

df[:2]
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1101S_1C_1Mstreet_11736334.0A+
1102S_1C_1Fstreet_21927332.5B+
df[df['Physics']=='A+']
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1101S_1C_1Mstreet_11736334.0A+
1203S_1C_2Mstreet_61605358.8A+
2203S_2C_2Mstreet_41559173.8A+

使用布尔值DataFrame进行索引,布尔值DataFrame可以是对表两只进行比较产生的。
这种索引方式在语法上更像是Numpy二维数组。

1.单级索引

1. 1 loc方法、iloc方法、[ ]操作符

最常用的索引方法可能就是这三类,其中iloc表示位置索引,loc表示标签索引,[ ]也具有很大的便利性,各有特点

1.1.1 loc方法(注意:所有在loc中使用的切片全部包含右端点!)

① 单行索引:

df.loc[1103]
School          S_1
Class           C_1
Gender            M
Address    street_2
Height          186
Weight           82
Math           87.2
Physics          B+
Name: 1103, dtype: object

② 多行索引:

df.loc[[1102,2304]]
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1102S_1C_1Fstreet_21927332.5B+
2304S_2C_3Fstreet_61648195.5A-
df.loc[1304:].head()
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1304S_1C_3Mstreet_21957085.2A
1305S_1C_3Fstreet_51876961.7B-
2101S_2C_1Mstreet_71748483.3C
2102S_2C_1Fstreet_61616150.6B+
2103S_2C_1Mstreet_41576152.5B-
df.loc[2402::-1].head()
SchoolClassGenderAddressHeightWeightMathPhysics
ID
2402S_2C_4Mstreet_71668248.7B
2401S_2C_4Fstreet_21926245.3A
2305S_2C_3Mstreet_41877348.9B
2304S_2C_3Fstreet_61648195.5A-
2303S_2C_3Fstreet_71909965.9C

③ 单列索引:

df.loc[:,'Height'].head()
ID
1101    173
1102    192
1103    186
1104    167
1105    159
Name: Height, dtype: int64

④ 多列索引:

df.loc[:,['Height','Math']].head()
HeightMath
ID
110117334.0
110219232.5
110318687.2
110416780.4
110515984.8
df.loc[:,'Height':'Math'].head()
HeightWeightMath
ID
11011736334.0
11021927332.5
11031868287.2
11041678180.4
11051596484.8

⑥ 函数式索引:

df.loc[lambda x:x['Gender']=='M'].head()
#loc中使用的函数,传入参数就是前面的df
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1101S_1C_1Mstreet_11736334.0A+
1103S_1C_1Mstreet_21868287.2B+
1201S_1C_2Mstreet_51886897.0A-
1203S_1C_2Mstreet_61605358.8A+
1301S_1C_3Mstreet_41616831.5B+
def f(x):
    return [1101,1103]
df.loc[f]
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1101S_1C_1Mstreet_11736334.0A+
1103S_1C_1Mstreet_21868287.2B+

⑦ 布尔索引

df.loc[df['Address'].isin(['street_7','street_4'])].head()
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1105S_1C_1Fstreet_41596484.8B+
1202S_1C_2Fstreet_41769463.5B-
1301S_1C_3Mstreet_41616831.5B+
1303S_1C_3Mstreet_71888249.7B
2101S_2C_1Mstreet_71748483.3C
df.loc[[True if i[-1]=='4' or i[-1]=='7' else False for i in df['Address'].values]].head()
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1105S_1C_1Fstreet_41596484.8B+
1202S_1C_2Fstreet_41769463.5B-
1301S_1C_3Mstreet_41616831.5B+
1303S_1C_3Mstreet_71888249.7B
2101S_2C_1Mstreet_71748483.3C

当然也有相对简便的写法:

df[df['Address'].str.contains('4|7')].head()

本质上说,loc中能传入的只有布尔列表和索引子集构成的列表,只要把握这个原则就很容易理解上面那些操作

1.1.2 iloc方法(注意与loc不同,切片右端点不包含)

  1. 单行索引:
df.iloc[3]
School          S_1
Class           C_1
Gender            F
Address    street_2
Height          167
Weight           81
Math           80.4
Physics          B-
Name: 1104, dtype: object
  1. 多行索引:
df.iloc[3:5]
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1104S_1C_1Fstreet_21678180.4B-
1105S_1C_1Fstreet_41596484.8B+

3.单列索引:

df.iloc[:,3].head()

ID
1101    street_1
1102    street_2
1103    street_2
1104    street_2
1105    street_4
Name: Address, dtype: object
  1. 多列索引:
df.iloc[:,7::-2].head()
PhysicsWeightAddressClass
ID
1101A+63street_1C_1
1102B+73street_2C_1
1103B+82street_2C_1
1104B-81street_2C_1
1105B+64street_4C_1
  1. 混合索引:
df.iloc[3::4,7::-2].head()
PhysicsWeightAddressClass
ID
1104B-81street_2C_1
1203A+53street_6C_2
1302A-57street_1C_3
2101C84street_7C_1
2105A81street_4C_1
  1. 函数式索引:
df.iloc[lambda x:[3]].head()
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1104S_1C_1Fstreet_21678180.4B-

由上所述,iloc中接收的参数只能为整数或整数列表,不能使用布尔索引

(c) [ ]操作符
如果不想陷入困境,请不要在行索引为浮点时使用 [ ]操作符,因为在Series中的浮点 [ ]并不是进行位置比较,而是值比较,非常特殊

  1. Series的 [ ]操作
    1.1 单元素索引:
s = pd.Series(df['Math'],index=df.index)
s[1101]
#使用的是索引标签
34.0

1.2 多行索引:

s[0:4]
#使用的是绝对位置的整数切片,与元素无关,这里容易混淆
ID
1101    34.0
1102    32.5
1103    87.2
1104    80.4
Name: Math, dtype: float64

1.3 函数式索引:

s[lambda x: x.index[16::-6]]
#注意使用lambda函数时,直接切片(如:s[lambda x: 16::-6])就报错,此时使用的不是绝对位置切片,而是元素切片,非常易错
ID
2102    50.6
1301    31.5
1105    84.8
Name: Math, dtype: float64

1.4 布尔索引:

s[s>80]
ID
1103    87.2
1104    80.4
1105    84.8
1201    97.0
1302    87.7
1304    85.2
2101    83.3
2205    85.4
2304    95.5
Name: Math, dtype: float64
  1. DataFrame的[ ]操作
    2.1 单行索引:
df[1:2]
#这里非常容易写成df['label'],会报错
#同Series使用了绝对位置切片
#如果想要获得某一个元素,可用如下get_loc方法:
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1102S_1C_1Fstreet_21927332.5B+
row = df.index.get_loc(1102)
df[row:row+1]
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1102S_1C_1Fstreet_21927332.5B+

2.2 多行索引:

#用切片,如果是选取指定的某几行,推荐使用loc,否则很可能报错
df[3:5]
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1104S_1C_1Fstreet_21678180.4B-
1105S_1C_1Fstreet_41596484.8B+

2.3 单列索引:

df['School'].head()
ID
1101    S_1
1102    S_1
1103    S_1
1104    S_1
1105    S_1
Name: School, dtype: object

2.4 多列索引:

df[['School','Math']].head()
SchoolMath
ID
1101S_134.0
1102S_132.5
1103S_187.2
1104S_180.4
1105S_184.8

2.5 函数式索引:

df[lambda x:['Math','Physics']].head()
MathPhysics
ID
110134.0A+
110232.5B+
110387.2B+
110480.4B-
110584.8B+

2.6 布尔索引:

df[df['Gender']=='F'].head()
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1102S_1C_1Fstreet_21927332.5B+
1104S_1C_1Fstreet_21678180.4B-
1105S_1C_1Fstreet_41596484.8B+
1202S_1C_2Fstreet_41769463.5B-
1204S_1C_2Fstreet_51626333.8B

一般来说,[]操作符常用于列选择或布尔选择,尽量避免行的选择.

1.2 布尔索引

1.2.1 布尔符号:’&’,’|’,’~’:分别代表和and,或or,取反not

df[(df['Gender']=='F')&(df['Address']=='street_2')].head()
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1102S_1C_1Fstreet_21927332.5B+
1104S_1C_1Fstreet_21678180.4B-
2401S_2C_4Fstreet_21926245.3A
2404S_2C_4Fstreet_21608467.7B
df[(df['Math']>85)|(df['Address']=='street_7')].head()
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1103S_1C_1Mstreet_21868287.2B+
1201S_1C_2Mstreet_51886897.0A-
1302S_1C_3Fstreet_11755787.7A-
1303S_1C_3Mstreet_71888249.7B
1304S_1C_3Mstreet_21957085.2A
df[~((df['Math']>75)|(df['Address']=='street_1'))].head()
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1102S_1C_1Fstreet_21927332.5B+
1202S_1C_2Fstreet_41769463.5B-
1203S_1C_2Mstreet_61605358.8A+
1204S_1C_2Fstreet_51626333.8B
1205S_1C_2Fstreet_61676368.4B-

loc和[]中相应位置都能使用布尔列表选择:

df.loc[df['Math']>60,(df[:8]['Address']=='street_6').values].head()
#如果不加values就会索引对齐发生错误,Pandas中的索引对齐是一个重要特征,很多时候非常使用
#但是若不加以留意,就会埋下隐患
Physics
ID
1103B+
1104B-
1105B+
1201A-
1202B-

1.2.2 isin方法

df[df['Address'].isin(['street_1','street_4'])&df['Physics'].isin(['A','A+'])]
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1101S_1C_1Mstreet_11736334.0A+
2105S_2C_1Mstreet_41708134.2A
2203S_2C_2Mstreet_41559173.8A+
#上面也可以用字典方式写:
df[df[['Address','Physics']].isin({'Address':['street_1','street_4'],'Physics':['A','A+']}).all(1)]
#all与&的思路是类似的,其中的1代表按照跨列方向判断是否全为True
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1101S_1C_1Mstreet_11736334.0A+
2105S_2C_1Mstreet_41708134.2A
2203S_2C_2Mstreet_41559173.8A+

1.2.3 快速标量索引

当只需要取一个元素时,at和iat方法能够提供更快的实现:

display(df.at[1101,'School'])
display(df.loc[1101,'School'])
display(df.iat[0,0])
display(df.iloc[0,0])
#可尝试去掉注释对比时间
#%timeit df.at[1101,'School']
#%timeit df.loc[1101,'School']
#%timeit df.iat[0,0]
#%timeit df.iloc[0,0]

'S_1'
'S_1'
'S_1'
'S_1'
#9.71 µs ± 488 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)
#20.7 µs ± 3.68 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)
#13 µs ± 2.53 µs per loop (mean ± std. dev. of 7 runs, 100000 loops each)
#16.3 µs ± 477 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)

1.2.4 区间索引

1.2.4.1利用interval_range方法
pd.interval_range(start=0,end=5)
#closed参数可选'left''right''both''neither',默认左开右闭
IntervalIndex([(0, 1], (1, 2], (2, 3], (3, 4], (4, 5]],
              closed='right',
              dtype='interval[int64]')
pd.interval_range(start=0,periods=8,freq=5)
#periods参数控制区间个数,freq控制步长
IntervalIndex([(0, 5], (5, 10], (10, 15], (15, 20], (20, 25], (25, 30], (30, 35], (35, 40]],
              closed='right',
              dtype='interval[int64]')
1.2.4.2 利用cut将数值列转为区间为元素的分类变量

例如统计数学成绩的区间情况:

math_interval = pd.cut(df['Math'],bins=[0,40,60,80,100])
#注意,如果没有类型转换,此时并不是区间类型,而是category类型
math_interval.head()
ID
1101      (0, 40]
1102      (0, 40]
1103    (80, 100]
1104    (80, 100]
1105    (80, 100]
Name: Math, dtype: category
Categories (4, interval[int64]): [(0, 40] < (40, 60] < (60, 80] < (80, 100]]

1.2.5 区间索引的选取

df_i = df.join(math_interval,rsuffix='_interval')[['Math','Math_interval']]\
            .reset_index().set_index('Math_interval')
df_i.head()
IDMath
Math_interval
(0, 40]110134.0
(0, 40]110232.5
(80, 100]110387.2
(80, 100]110480.4
(80, 100]110584.8
df_i.loc[65].head()
#包含该值就会被选中
IDMath
Math_interval
(60, 80]120263.5
(60, 80]120568.4
(60, 80]130561.7
(60, 80]210472.2
(60, 80]220268.5
df_i.loc[[65,90]].head()
IDMath
Math_interval
(60, 80]120263.5
(60, 80]120568.4
(60, 80]130561.7
(60, 80]210472.2
(60, 80]220268.5

如果想要选取某个区间,先要把分类变量转为区间变量,再使用overlap方法:

df_i[df_i.index.astype('interval').overlaps(pd.Interval(70, 85))].head()
IDMath
Math_interval
(80, 100]110387.2
(80, 100]110480.4
(80, 100]110584.8
(80, 100]120197.0
(60, 80]120263.5

2 多级索引

2.1 创建多级索引

2.1.1 通过from_tuple或from_arrays

  1. 直接创建元组
tuples = [('A','a'),('A','b'),('B','a'),('B','b')]
mul_index = pd.MultiIndex.from_tuples(tuples, names=('Upper', 'Lower'))
mul_index
MultiIndex([('A', 'a'),
            ('A', 'b'),
            ('B', 'a'),
            ('B', 'b')],
           names=['Upper', 'Lower'])
pd.DataFrame({'Score':['perfect','good','fair','bad']},index=mul_index)
Score
UpperLower
Aaperfect
bgood
Bafair
bbad
  1. 利用zip创建元组:
L1 = list('AABB')
L2 = list('abab')
tuples = list(zip(L1,L2))
mul_index = pd.MultiIndex.from_tuples(tuples, names=('Upper', 'Lower'))
pd.DataFrame({'Score':['perfect','good','fair','bad']},index=mul_index)
Score
UpperLower
Aaperfect
bgood
Bafair
bbad
  1. 通过Array创建
arrays = [['A','a'],['A','b'],['B','a'],['B','b']]
mul_index = pd.MultiIndex.from_tuples(arrays, names=('Upper', 'Lower'))
pd.DataFrame({'Score':['perfect','good','fair','bad']},index=mul_index)
Score
UpperLower
Aaperfect
bgood
Bafair
bbad
mul_index
#由此看出内部自动转成元组
MultiIndex([('A', 'a'),
            ('A', 'b'),
            ('B', 'a'),
            ('B', 'b')],
           names=['Upper', 'Lower'])

2.1.2 通过from_product

L1 = ['A','B']
L2 = ['a','b']
pd.MultiIndex.from_product([L1,L2],names=('Upper', 'Lower'))
#两两相乘
MultiIndex([('A', 'a'),
            ('A', 'b'),
            ('B', 'a'),
            ('B', 'b')],
           names=['Upper', 'Lower'])

2.1.3 指定df中的列创建(set_index方法)

df_using_mul = df.set_index(['Class','Address'])
df_using_mul.head()
SchoolGenderHeightWeightMathPhysics
ClassAddress
C_1street_1S_1M1736334.0A+
street_2S_1F1927332.5B+
street_2S_1M1868287.2B+
street_2S_1F1678180.4B-
street_4S_1F1596484.8B+

2.2 多层索引切片

df_using_mul.head()
SchoolGenderHeightWeightMathPhysics
ClassAddress
C_1street_1S_1M1736334.0A+
street_2S_1F1927332.5B+
street_2S_1M1868287.2B+
street_2S_1F1678180.4B-
street_4S_1F1596484.8B+

2.2.1 一般切片

#df_using_mul.loc['C_2','street_5']
#当索引不排序时,单个索引会报出性能警告
#df_using_mul.index.is_lexsorted()
#该函数检查是否排序
df_using_mul.sort_index().loc['C_2','street_5']
#df_using_mul.sort_index().index.is_lexsorted()
SchoolGenderHeightWeightMathPhysics
ClassAddress
C_2street_5S_1M1886897.0A-
street_5S_1F1626333.8B
street_5S_2M19310039.1B
#df_using_mul.loc[('C_2','street_5'):] 报错
#当不排序时,不能使用多层切片
df_using_mul.sort_index().loc[('C_2','street_6'):('C_3','street_4')]
#注意此处由于使用了loc,因此仍然包含右端点
SchoolGenderHeightWeightMathPhysics
ClassAddress
C_2street_6S_1M1605358.8A+
street_6S_1F1676368.4B-
street_7S_2F1947768.5B+
street_7S_2F1837685.4B
C_3street_1S_1F1755787.7A-
street_2S_1M1957085.2A
street_4S_1M1616831.5B+
street_4S_2F1577872.3B+
street_4S_2M1877348.9B
df_using_mul.sort_index().loc[('C_2','street_7'):'C_3'].head()
#非元组也是合法的,表示选中该层所有元素
SchoolGenderHeightWeightMathPhysics
ClassAddress
C_2street_7S_2F1947768.5B+
street_7S_2F1837685.4B
C_3street_1S_1F1755787.7A-
street_2S_1M1957085.2A
street_4S_1M1616831.5B+

2.2.2 第一类特殊情况:由元组构成列表

df_using_mul.sort_index().loc[[('C_2','street_7'),('C_3','street_2')]]
#表示选出某几个元素,精确到最内层索引
SchoolGenderHeightWeightMathPhysics
ClassAddress
C_2street_7S_2F1947768.5B+
street_7S_2F1837685.4B
C_3street_2S_1M1957085.2A

2.2.3 第二类特殊情况:由列表构成元组

df_using_mul.sort_index().loc[(['C_2','C_3'],['street_4','street_7']),:]
#选出第一层在‘C_2’和'C_3'中且第二层在'street_4'和'street_7'中的行
SchoolGenderHeightWeightMathPhysics
ClassAddress
C_2street_4S_1F1769463.5B-
street_4S_2M1559173.8A+
street_7S_2F1947768.5B+
street_7S_2F1837685.4B
C_3street_4S_1M1616831.5B+
street_4S_2F1577872.3B+
street_4S_2M1877348.9B
street_7S_1M1888249.7B
street_7S_2F1909965.9C

2.3 多层索引中的slice对象

L1,L2 = ['A','B','C'],['a','b','c']
mul_index1 = pd.MultiIndex.from_product([L1,L2],names=('Upper', 'Lower'))
L3,L4 = ['D','E','F'],['d','e','f']
mul_index2 = pd.MultiIndex.from_product([L3,L4],names=('Big', 'Small'))
df_s = pd.DataFrame(np.random.rand(9,9),index=mul_index1,columns=mul_index2)
df_s
BigDEF
Smalldefdefdef
UpperLower
Aa0.2948490.8089440.2868610.0798010.4980020.9744760.2616530.7923290.671085
b0.0382810.8022290.6892210.2584380.5265830.5270280.6622870.4242680.312776
c0.3602620.0969730.4918840.8435970.0401540.5636630.0659390.3423670.928119
Ba0.4833220.2261180.6958740.8040360.0566070.4055350.9608900.4095570.044850
b0.9027730.8896560.2129780.9863620.7338430.2347460.2669320.3606930.856747
c0.2040790.4243080.8786920.0278830.6048380.2922230.8144610.6336490.195861
Ca0.7451680.0652480.4088810.6473610.0423350.2456390.3513570.9456640.546389
b0.6823880.9487410.0726720.1797050.7753700.5655600.9357420.1496200.806942
c0.0798930.0317650.4253900.5578840.8277650.6502210.5197170.7080330.128069
idx=pd.IndexSlice
df_s.loc[idx['B':,df_s['D']['d']>0.3],idx[df_s.sum()>4]]
#df_s.sum()默认为对列求和,因此返回一个长度为9的数值列表
BigDEF
Smalldefefe
UpperLower
Ba0.88980.3646980.3477620.9750530.4943570.600218
b0.8311980.9223690.1213230.1040720.9652470.951544
Ca0.4695550.2023890.3603230.7897730.6201750.284863
c0.6843780.4940560.9954410.7380330.1985390.770315

2.4 索引层的交换

2.4.1 swaplevel方法(两层交换)

df_using_mul.head()
ClassAddress
C_1street_1S_1M1736334A+
street_2S_1F1927332.5B+
street_2S_1M1868287.2B+
street_2S_1F1678180.4B-
street_4S_1F1596484.8B+
df_using_mul.swaplevel(i=1,j=0,axis=0).sort_index().head()
SchoolGenderHeightWeightMathPhysics
AddressClass
street_1C_1S_1M1736334A+
C_2S_2M1757447.2B-
C_3S_1F1755787.7A-
street_2C_1S_1F1927332.5B+
C_1S_1M1868287.2B+

2.4.2 reorder_levels方法(多层交换)

df_muls = df.set_index(['School','Class','Address'])
df_muls.head()
GenderHeightWeightMathPhysics
SchoolClassAddress
S_1C_1street_1M1736334A+
street_2F1927332.5B+
street_2M1868287.2B+
street_2F1678180.4B-
street_4F1596484.8B+
df_muls.reorder_levels([2,0,1],axis=0).sort_index().head()
GenderHeightWeightMathPhysics
AddressSchoolClass
street_1S_1C_1M1736334A+
C_3F1755787.7A-
S_2C_2M1757447.2B-
street_2S_1C_1F1927332.5B+
C_1M1868287.2B+
#如果索引有name,可以直接使用name
df_muls.reorder_levels(['Address','School','Class'],axis=0).sort_index().head()
GenderHeightWeightMathPhysics
AddressSchoolClass
street_1S_1C_1M1736334A+
C_3F1755787.7A-
S_2C_2M1757447.2B-
street_2S_1C_1F1927332.5B+
C_1M1868287.2B+

3 索引设定

3.1 index_col参数

index_col是read_csv中的一个参数,而不是某一个方法:

pd.read_csv('data/table.csv',index_col=['Address','School']).head()
ClassIDGenderHeightWeightMathPhysics
AddressSchool
street_1S_1C_11101M1736334A+
street_2S_1C_11102F1927332.5B+
S_1C_11103M1868287.2B+
S_1C_11104F1678180.4B-
street_4S_1C_11105F1596484.8B+

3.2 reindex和reindex_like

reindex是指重新索引,它的重要特性在于索引对齐,很多时候用于重新排序

df.head()
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1101S_1C_1Mstreet_11736334.0A+
1102S_1C_1Fstreet_21927332.5B+
1103S_1C_1Mstreet_21868287.2B+
1104S_1C_1Fstreet_21678180.4B-
1105S_1C_1Fstreet_41596484.8B+
df.reindex(index=[1101,1203,1206,2402])
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1101S_1C_1Mstreet_1173.063.034.0A+
1203S_1C_2Mstreet_6160.053.058.8A+
1206NaNNaNNaNNaNNaNNaNNaNNaN
2402S_2C_4Mstreet_7166.082.048.7B
df.reindex(columns=['Height','Gender','Average']).head()
HeightGenderAverage
ID
1101173MNaN
1102192FNaN
1103186MNaN
1104167FNaN
1105159FNaN

可以选择缺失值的填充方法:fill_value和method(bfill/ffill/nearest),其中method参数必须索引单调。

df.reindex(index=[1101,1203,1206,2402],method='bfill')
#bfill表示用所在索引1206的后一个有效行填充,ffill为前一个有效行,nearest是指最近的
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1101S_1C_1Mstreet_11736334.0A+
1203S_1C_2Mstreet_61605358.8A+
1206S_1C_3Mstreet_41616831.5B+
2402S_2C_4Mstreet_71668248.7B
df.reindex(index=[1101,1203,1206,2402],method='nearest')
#数值上1205比1301更接近1206,因此用前者填充
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1101S_1C_1Mstreet_11736334.0A+
1203S_1C_2Mstreet_61605358.8A+
1206S_1C_2Fstreet_61676368.4B-
2402S_2C_4Mstreet_71668248.7B

reindex_like的作用为生成一个横纵索引完全与参数列表一致的DataFrame,数据使用被调用的表

df_temp = pd.DataFrame({'Weight':np.zeros(5),
                        'Height':np.zeros(5),
                        'ID':[1101,1104,1103,1106,1102]}).set_index('ID')
df_temp.reindex_like(df[0:5][['Weight','Height']])
WeightHeight
ID
11010.00.0
11020.00.0
11030.00.0
11040.00.0
1105NaNNaN

如果df_temp单调还可以使用method参数:

df_temp = pd.DataFrame({'Weight':range(5),
                        'Height':range(5),
                        'ID':[1101,1104,1103,1106,1102]}).set_index('ID').sort_index()
df_temp.reindex_like(df[0:5][['Weight','Height']],method='bfill')
#可以自行检验这里的1105的值是否是由bfill规则填充
WeightHeight
ID
110100
110244
110322
110411
110533

3.3 set_index和reset_index

et_index:从字面意思看,就是将某些列作为索引
使用表内列作为索引:

df.head()
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1101S_1C_1Mstreet_11736334.0A+
1102S_1C_1Fstreet_21927332.5B+
1103S_1C_1Mstreet_21868287.2B+
1104S_1C_1Fstreet_21678180.4B-
1105S_1C_1Fstreet_41596484.8B+
df.set_index('Class').head()
SchoolGenderAddressHeightWeightMathPhysics
Class
C_1S_1Mstreet_11736334.0A+
C_1S_1Fstreet_21927332.5B+
C_1S_1Mstreet_21868287.2B+
C_1S_1Fstreet_21678180.4B-
C_1S_1Fstreet_41596484.8B+

利用append参数可以将当前索引维持不变

df.set_index('Class',append=True).head()
SchoolGenderAddressHeightWeightMathPhysics
IDClass
1101C_1S_1Mstreet_11736334.0A+
1102C_1S_1Fstreet_21927332.5B+
1103C_1S_1Mstreet_21868287.2B+
1104C_1S_1Fstreet_21678180.4B-
1105C_1S_1Fstreet_41596484.8

当使用与表长相同的列作为索引(需要先转化为Series,否则报错):

df.set_index(pd.Series(range(df.shape[0]))).head()
SchoolClassGenderAddressHeightWeightMathPhysics
0S_1C_1Mstreet_11736334.0A+
1S_1C_1Fstreet_21927332.5B+
2S_1C_1Mstreet_21868287.2B+
3S_1C_1Fstreet_21678180.4B-
4S_1C_1Fstreet_41596484.8B+

可以直接添加多级索引:

df.set_index([pd.Series(range(df.shape[0])),pd.Series(np.ones(df.shape[0]))]).head()
SchoolClassGenderAddressHeightWeightMathPhysics
01.0S_1C_1Mstreet_11736334.0A+
11.0S_1C_1Fstreet_21927332.5B+
21.0S_1C_1Mstreet_21868287.2B+
31.0S_1C_1Fstreet_21678180.4B-
41.0S_1C_1Fstreet_41596484.8B+

reset_index方法,它的主要功能是将索引重置
默认状态直接恢复到自然数索引:

df.reset_index().head()
IDSchoolClassGenderAddressHeightWeightMathPhysics
01101S_1C_1Mstreet_11736334.0A+
11102S_1C_1Fstreet_21927332.5B+
21103S_1C_1Mstreet_21868287.2B+
31104S_1C_1Fstreet_21678180.4B-
41105S_1C_1Fstreet_41596484.8B+

用level参数指定哪一层被reset,用col_level参数指定set到哪一层:

L1,L2 = ['A','B','C'],['a','b','c']
mul_index1 = pd.MultiIndex.from_product([L1,L2],names=('Upper', 'Lower'))
L3,L4 = ['D','E','F'],['d','e','f']
mul_index2 = pd.MultiIndex.from_product([L3,L4],names=('Big', 'Small'))
df_temp = pd.DataFrame(np.random.rand(9,9),index=mul_index1,columns=mul_index2)
df_temp.head()
BigDEF
Smalldefdefdef
UpperLower
Aa0.4151950.3998410.4320310.2687360.1485710.9723630.894480.156690.462923
b0.4905080.1505710.100150.7726130.8474220.3200510.5506750.0694990.191721
c0.1935150.0695480.7202090.1806380.9888390.1116570.6491080.5297430.739764
Ba0.9563250.1396720.6310070.7043350.1048280.2623780.6071940.7256910.096959
b0.4656110.2886480.2367060.7061290.2983130.3423650.5433230.4902580.027482
df_temp1 = df_temp.reset_index(level=1,col_level=1)
df_temp1.head()
BigDEF
SmallLowerdefdefdef
Upper
Aa0.4557720.5683830.366010.1428340.2327440.0459310.854950.0823190.80917
Ab0.4468740.4592030.492440.8910930.6194260.1271370.1361370.1111080.424913
Ac0.67190.1951270.1757040.9404910.0178820.1958290.8449490.3358650.55247
Ba0.1146340.9737960.8471490.6060020.9390.7733610.528390.7008320.516844
Bb0.6836240.5445240.9040160.5188690.8396970.8348560.8748630.7667190.598977
df_temp1.columns
#看到的确插入了level2
MultiIndex([( '', 'Lower'),
            ('D',     'd'),
            ('D',     'e'),
            ('D',     'f'),
            ('E',     'd'),
            ('E',     'e'),
            ('E',     'f'),
            ('F',     'd'),
            ('F',     'e'),
            ('F',     'f')],
           names=['Big', 'Small'])
df_temp1.index
#最内层索引被移出
Index(['A', 'A', 'A', 'B', 'B', 'B', 'C', 'C', 'C'], dtype='object', name='Upper')

3.4 rename_axis和rename

rename_axis是针对多级索引的方法,作用是修改某一层的索引名,而不是索引标签

df_temp.rename_axis(index={'Lower':'LowerLower'},columns={'Big':'BigBig'})
BigBigDEF
Smalldefdefdef
UpperLowerLower
Aa0.0251690.1131440.1182850.7750660.8014920.4973750.9586160.6853570.549634
b0.8374980.3378930.2316330.8316860.7160310.6260490.6190130.8773520.310406
c0.1157370.1115870.9225710.2584030.646570.9368690.8937630.0147680.533456
Ba0.0244060.159180.5307970.3127140.2304860.3102520.6950730.1955270.302385
b0.4530450.4416130.3271580.559870.9513310.9034110.4061680.5866510.284886
c0.660420.7198560.8603040.0646050.3311110.1636860.9207320.4096730.219974
Ca0.6840490.9007080.8708610.0447540.0264870.1460950.6489230.1871750.704501
b0.2374860.68410.2556660.1662650.4017160.0408350.9148650.6963180.653498
c0.5595610.0444970.8030690.3740560.6889140.560570.291540.7050210.60011

rename方法用于修改列或者行索引标签,而不是索引名:

df_temp.rename(index={'A':'T'},columns={'e':'changed_e'}).head()
BigDEF
Smalldchanged_efdchanged_efdchanged_ef
UpperLower
Ta0.9679450.100470.5926430.1924160.9902230.0988810.1986540.4228010.304565
b0.3261640.2895460.8291210.7613380.0087720.7384610.3406970.5137270.385448
c0.4755340.5131950.8258450.6657170.3596060.2072180.556010.095650.880627
Ba0.757570.8726310.081050.521730.78180.840370.6492060.5773880.0943
b0.9911140.0331760.3923720.5906030.6079970.1717890.0363820.1065610.893742

4 常用索引型函数

4.1 where函数

当对条件为False的单元进行填充:

df.where(df['Gender']=='M').head()
#不满足条件的行全部被设置为NaN
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1101S_1C_1Mstreet_1173.063.034.0A+
1102NaNNaNNaNNaNNaNNaNNaNNaN
1103S_1C_1Mstreet_2186.082.087.2B+
1104NaNNaNNaNNaNNaNNaNNaNNaN
1105NaNNaNNaNNaNNaNNaNNaNNaN

通过这种方法筛选结果和[]操作符的结果完全一致:

df.where(df['Gender']=='M').dropna().head()
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1101S_1C_1Mstreet_1173.063.034.0A+
1103S_1C_1Mstreet_2186.082.087.2B+
1201S_1C_2Mstreet_5188.068.097.0A-
1203S_1C_2Mstreet_6160.053.058.8A+
1301S_1C_3Mstreet_4161.068.031.5B+

第一个参数为布尔条件,第二个参数为填充值:

df.where(df['Gender']=='M',np.random.rand(df.shape[0],df.shape[1])).head()
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1101S_1C_1Mstreet_1173.00000063.00000034.000000A+
11020.01524670.7084440.9171990.3021850.6896430.0101260.7246360.895387
1103S_1C_1Mstreet_2186.00000082.00000087.200000B+
11040.3691950.4592110.4641910.9644860.3657970.1276020.5014960.0287754
11050.8122320.9996340.8257820.2856920.3401970.0839820.7923100.133054

4.2 mask函数

mask函数与where功能上相反,其余完全一致,即对条件为True的单元进行填充

df.mask(df['Gender']=='M').dropna().head()
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1102S_1C_1Fstreet_2192.073.032.5B+
1104S_1C_1Fstreet_2167.081.080.4B-
1105S_1C_1Fstreet_4159.064.084.8B+
1202S_1C_2Fstreet_4176.094.063.5B-
1204S_1C_2Fstreet_5162.063.033.8B
df.mask(df['Gender']=='M',np.random.rand(df.shape[0],df.shape[1])).head()
SchoolClassGenderAddressHeightWeightMathPhysics
ID
11010.2739620.250280.5874710.9772060.4424030.3194600.4609910.842498
1102S_1C_1Fstreet_2192.00000073.00000032.500000B+
11030.4366740.7415240.469960.6886030.9382410.5318110.7943520.17495
1104S_1C_1Fstreet_2167.00000081.00000080.400000B-
1105S_1C_1Fstreet_4159.00000064.00000084.800000B+

4.3 query函数

query函数中的布尔表达式中,下面的符号都是合法的:行列索引名、字符串、and/not/or/&/|/~/not in/in/==/!=、四则运算符

df.query('(Address in ["street_6","street_7"])&(Weight>(70+10))&(ID in [1303,2304,2402])')
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1303S_1C_3Mstreet_71888249.7B
2304S_2C_3Fstreet_61648195.5A-
2402S_2C_4Mstreet_71668248.7B

5 重复元素处理

5.1 duplicated方法

该方法返回了是否重复的布尔列表

df.duplicated('Class').head()
ID
1101    False
1102     True
1103     True
1104     True
1105     True
dtype: bool

可选参数keep默认为first,即首次出现设为不重复,若为last,则最后一次设为不重复,若为False,则所有重复项为False

df.duplicated('Class',keep='last').tail()
ID
2401     True
2402     True
2403     True
2404     True
2405    False
dtype: bool
df.duplicated('Class',keep=False).head()
ID
1101    True
1102    True
1103    True
1104    True
1105    True
dtype: bool

5.2 drop_duplicates方法

剔除重复项

df.drop_duplicates('Class')
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1101S_1C_1Mstreet_11736334.0A+
1201S_1C_2Mstreet_51886897.0A-
1301S_1C_3Mstreet_41616831.5B+
2401S_2C_4Fstreet_21926245.3A

参数与duplicate函数类似:

df.drop_duplicates('Class',keep='last')
SchoolClassGenderAddressHeightWeightMathPhysics
ID
2105S_2C_1Mstreet_41708134.2A
2205S_2C_2Fstreet_71837685.4B
2305S_2C_3Mstreet_41877348.9B
2405S_2C_4Fstreet_61935447.6B

在传入多列时等价于将多列共同视作一个多级索引,比较重复项:

df.drop_duplicates(['School','Class'])
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1101S_1C_1Mstreet_11736334.0A+
1201S_1C_2Mstreet_51886897.0A-
1301S_1C_3Mstreet_41616831.5B+
2101S_2C_1Mstreet_71748483.3C
2201S_2C_2Mstreet_519310039.1B
2301S_2C_3Fstreet_41577872.3B+
2401S_2C_4Fstreet_21926245.3A

6 抽样函数

这里的抽样函数指的就是sample函数

6.1 n为样本量

df.sample(n=5)
SchoolClassGenderAddressHeightWeightMathPhysics
ID
2103S_2C_1Mstreet_41576152.5B-
1102S_1C_1Fstreet_21927332.5B+
1301S_1C_3Mstreet_41616831.5B+
1304S_1C_3Mstreet_21957085.2A
1105S_1C_1Fstreet_41596484.8B+

6.2 frac为抽样比

df.sample(frac=0.05)
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1105S_1C_1Fstreet_41596484.8B+
2402S_2C_4Mstreet_71668248.7B

6.3 replace为是否放回

df.sample(n=df.shape[0],replace=True).head()
SchoolClassGenderAddressHeightWeightMathPhysics
ID
2403S_2C_4Fstreet_61586059.7B+
2404S_2C_4Fstreet_21608467.7B
2405S_2C_4Fstreet_61935447.6B
2303S_2C_3Fstreet_71909965.9C
1203S_1C_2Mstreet_61605358.8A+
df.sample(n=35,replace=True).index.is_unique
False

6.4 axis为抽样维度,默认为0,即抽行

df.sample(n=3,axis=1).head()
AddressWeightSchool
ID
1101street_163S_1
1102street_273S_1
1103street_282S_1
1104street_281S_1
1105street_464S_1

6.5 weights为样本权重,自动归一化

df.sample(n=3,weights=np.random.rand(df.shape[0])).head()
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1302S_1C_3Fstreet_11755787.7A-
1305S_1C_3Fstreet_51876961.7B-
2404S_2C_4Fstreet_21608467.7B
#以某一列为权重,这在抽样理论中很常见
df.sample(n=3,weights=df['Math']).head()
SchoolClassGenderAddressHeightWeightMathPhysics
ID
1305S_1C_3Fstreet_51876961.7B-
2103S_2C_1Mstreet_41576152.5B-
2105S_2C_1Mstreet_41708134.2A
  • 3
    点赞
  • 11
    收藏
    觉得还不错? 一键收藏
  • 0
    评论
这篇笔记主要介绍了Pandas模块的基本操作和使用方法。PandasPython一个用于数据分析和处理的常用库,提供了高效的数据结构和数据分析工具,是进行数据处理和数据挖掘的重要工具之一。 一、Pandas数据结构 Pandas主要有两种数据结构:Series和DataFrame。 1. Series Series是一种类似于一维数组的对象,由一组数据和一组与之相关的标签(即索引)组成。Series的创建方式如下: ```python import pandas as pd # 通过列表创建Series s = pd.Series([1, 3, 5, np.nan, 6, 8]) # 通过字典创建Series s = pd.Series({'a': 1, 'b': 2, 'c': 3}) ``` 2. DataFrame DataFrame是一种二维表格数据结构,由一组数据和一组行索引和列索引组成。DataFrame的创建方式有很多种,最常用的是通过字典创建。例如: ```python import pandas as pd data = {'name': ['Tom', 'Jerry', 'Mike'], 'age': [18, 20, 22], 'gender': ['M', 'M', 'F']} df = pd.DataFrame(data) ``` 二、Pandas的基本操作 1. 数据读取 Pandas可以读取多种格式的数据文件,如CSV、Excel、SQL等。常用的读取CSV文件的方式如下: ```python import pandas as pd df = pd.read_csv('data.csv') ``` 2. 数据预处理 数据预处理是数据挖掘非常重要的一部分,Pandas提供了很多方便的函数和方法来进行数据清洗和转换。常用的数据预处理函数和方法有: - 处理缺失值 ```python # 判断是否存在缺失值 df.isnull() # 删除缺失值 df.dropna() # 填充缺失值 df.fillna(value) ``` - 处理重复值 ```python # 删除重复值 df.drop_duplicates() ``` - 数据转换 ```python # 数据类型转换 df.astype() # 数据替换 df.replace() ``` 3. 数据分析 Pandas提供了各种数据分析和处理的方法和函数,常用的包括: - 统计函数 ```python # 计算平均值 df.mean() # 计算标准差 df.std() # 计算最大值和最小值 df.max(), df.min() ``` - 排序 ```python # 按照某列排序 df.sort_values(by='column_name') ``` - 数据聚合 ```python # 对某列数据进行分组求和 df.groupby('column_name').sum() ``` 以上是Pandas模块的基础内容,还有很多高级用法和技巧需要进一步学习和掌握。

“相关推荐”对你有帮助么?

  • 非常没帮助
  • 没帮助
  • 一般
  • 有帮助
  • 非常有帮助
提交
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值