python基础知识之索引与切片详解

2022-11-13 08:11:57 切片 详解 基础知识

基本索引

In [4]: sentence = 'You are a nice girl'In [5]: L = sentence.split()In [6]: LOut[6]: ['You', 'are', 'a', 'nice', 'girl']

# 从0开始索引In [7]: L[2]Out[7]: 'a'

# 负数索引,从列表右侧开始计数In [8]: L[-2]Out[8]: 'nice'

# -1表示列表最后一项In [9]: L[-1]Out[9]: 'girl'

# 当正整数索引超过返回时In [10]: L[100]---------------------------------------------------------------------------IndexError                                Traceback (most recent call last)
<ipython-input-10-78da2f882365> in <module>()----> 1 L[100]IndexError: list index out of range# 当负整数索引超过返回时In [11]: L[-100]---------------------------------------------------------------------------IndexError                                Traceback (most recent call last)
<iPython-input-11-46b47b0ecb55> in <module>()----> 1 L[-100]IndexError: list index out of range# slice 索引In [193]: sl = slice(0,-1,1)In [194]: L[sl]Out[194]: ['You', 'are', 'a', 'nice']In [199]: sl = slice(0,100)In [200]: L[sl]Out[200]: ['You', 'are', 'a', 'nice', 'girl']

嵌套索引

In [14]: L = [[1,2,3],{'I':'You are a nice girl','She':'Thank you!'},(11,22),'My name is Kyles']

In [15]: L
Out[15]:
[[1, 2, 3],
 {'I': 'You are a nice girl', 'She': 'Thank you!'},
 (11, 22),
 'My name is Kyles']# 索引第1项,索引为0In [16]: L[0]
Out[16]: [1, 2, 3]# 索引第1项的第2子项In [17]: L[0][1]
Out[17]: 2# 索引第2项词典In [18]: L[1]
Out[18]: {'I': 'You are a nice girl', 'She': 'Thank you!'}# 索引第2项词典的 “She”In [19]: L[1]['She']
Out[19]: 'Thank you!'# 索引第3项In [20]: L[2]
Out[20]: (11, 22)# 索引第3项,第一个元组In [22]: L[2][0]
Out[22]: 11# 索引第4项In [23]: L[3]
Out[23]: 'My name is Kyles'# 索引第4项,前3个字符In [24]: L[3][:3]
Out[24]: 'My '

切片

# 切片选择,从1到列表末尾In [13]: L[1:]Out[13]: ['are', 'a', 'nice', 'girl']# 负数索引,选取列表后两项In [28]: L[-2:]Out[28]: ['nice', 'girl']# 异常测试,这里没有报错!In [29]: L[-100:]Out[29]: ['You', 'are', 'a', 'nice', 'girl']# 返回空In [30]: L[-100:-200]Out[30]: []# 正向索引In [32]: L[-100:3]Out[32]: ['You', 'are', 'a']# 返回空In [33]: L[-1:3]Out[33]: []# 返回空In [41]: L[0:0]Out[41]: []

看似简单的索引,有的人不以为然,我们这里采用精准的数字索引,很容易排查错误。若索引是经过计算出的一个变量,就千万要小心了,否则失之毫厘差之千里。

numpy.array 索引 一维

In [34]: import numpy as npIn [35]: arr = np.arange(10)In [36]: arrOut[36]: array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])In [40]: arr.shapeOut[40]: (10,)# [0,1) In [37]: arr[0:1]Out[37]: array([0])# [0,0) In [38]: arr[0:0]Out[38]: array([], dtype=int32)# 右侧超出范围之后In [42]: arr[:1000]Out[42]: array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])# 左侧超出之后In [43]: arr[-100:1000]Out[43]: array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])# 两侧都超出In [44]: arr[100:101]Out[44]: array([], dtype=int32)# []In [45]: arr[-100:-2]Out[45]: array([0, 1, 2, 3, 4, 5, 6, 7])# []In [46]: arr[-100:-50]Out[46]: array([], dtype=int32)

numpy.array 索引 二维

In [49]: arr = np.arange(15).reshape(3,5)

In [50]: arr
Out[50]:
array([[ 0,  1,  2,  3,  4],
       [ 5,  6,  7,  8,  9],
       [10, 11, 12, 13, 14]])

In [51]: arr.shape
Out[51]: (3, 5)

# axis = 0 增长的方向
In [52]: arr[0]
Out[52]: array([0, 1, 2, 3, 4])

# 选取第2行
In [53]: arr[1]
Out[53]: array([5, 6, 7, 8, 9])

# axis = 1 增长的方向,选取每一行的第1列
In [54]: arr[:,0]
Out[54]: array([ 0,  5, 10])

# axis = 1 增长的方向,选取每一行的第2列
In [55]: arr[:,1]
Out[55]: array([ 1,  6, 11])


# 选取每一行的第1,2列
In [56]: arr[:,0:2]
Out[56]:
array([[ 0,  1],
       [ 5,  6],
       [10, 11]])

# 右侧超出范围之后
In [57]: arr[:,0:100]
Out[57]:
array([[ 0,  1,  2,  3,  4],
       [ 5,  6,  7,  8,  9],
       [10, 11, 12, 13, 14]])

# 左侧超出范围之后
In [62]: arr[:,-10:2]
Out[62]:
array([[ 0,  1],
       [ 5,  6],
       [10, 11]])

# []
In [58]: arr[:,0:0]
Out[58]: array([], shape=(3, 0), dtype=int32)

# []
In [59]: arr[0:0,0:1]
Out[59]: array([], shape=(0, 1), dtype=int32)

# 异常
In [63]: arr[:,-10]---------------------------------------------------------------------------IndexError                                Traceback (most recent call last)
<ipython-input-63-2ffa6627dc7f> in <module>()----> 1 arr[:,-10]IndexError: index -10 is out of bounds for axis 1 with size 5

numpy.array 索引 三维…N维

In [67]: import numpy as np

In [68]: arr = np.arange(30).reshape(2,3,5)

In [69]: arr
Out[69]:
array([[[ 0,  1,  2,  3,  4],
        [ 5,  6,  7,  8,  9],
        [10, 11, 12, 13, 14]],       [[15, 16, 17, 18, 19],
        [20, 21, 22, 23, 24],
        [25, 26, 27, 28, 29]]])

# 根据 axis = 0 选取
In [70]: arr[0]
Out[70]:
array([[ 0,  1,  2,  3,  4],
       [ 5,  6,  7,  8,  9],
       [10, 11, 12, 13, 14]])

In [71]: arr[1]
Out[71]:
array([[15, 16, 17, 18, 19],
       [20, 21, 22, 23, 24],
       [25, 26, 27, 28, 29]])

# 根据 axis = 1 选取
In [72]: arr[:,0]
Out[72]:
array([[ 0,  1,  2,  3,  4],
       [15, 16, 17, 18, 19]])

In [73]: arr[:,1]
Out[73]:
array([[ 5,  6,  7,  8,  9],
       [20, 21, 22, 23, 24]])

# 异常指出 axis = 1 超出范围
In [74]: arr[:,4]---------------------------------------------------------------------------IndexError                                Traceback (most recent call last)
<ipython-input-74-9d489478e7c7> in <module>()----> 1 arr[:,4]IndexError: index 4 is out of bounds for axis 1 with size 3  # 根据 axis = 2 选取
In [75]: arr[:,:,0]
Out[75]:
array([[ 0,  5, 10],
       [15, 20, 25]])

# 降维
In [76]: arr[:,:,0].shape
Out[76]: (2, 3)

In [78]: arr[:,:,0:2]
Out[78]:
array([[[ 0,  1],
        [ 5,  6],
        [10, 11]],       [[15, 16],
        [20, 21],
        [25, 26]]])

In [79]: arr[:,:,0:2].shape
Out[79]: (2, 3, 2)

# 左/右侧超出范围
In [81]: arr[:,:,0:100]
Out[81]:
array([[[ 0,  1,  2,  3,  4],
        [ 5,  6,  7,  8,  9],
        [10, 11, 12, 13, 14]],       [[15, 16, 17, 18, 19],
        [20, 21, 22, 23, 24],
        [25, 26, 27, 28, 29]]])

# 异常 axis = 0In [82]: arr[100,:,0:100]---------------------------------------------------------------------------IndexError                                Traceback (most recent call last)
<ipython-input-82-21efcc74439d> in <module>()----> 1 arr[100,:,0:100]IndexError: index 100 is out of bounds for axis 0 with size 2

pandas Series 索引

In [84]: s = pd.Series(['You','are','a','nice','girl'])In [85]: sOut[85]:0     You1     are2       a3    nice4    girl
dtype: object# 按照索引选择In [86]: s[0]Out[86]: 'You'# []In [87]: s[0:0]Out[87]: Series([], dtype: object)In [88]: s[0:-1]Out[88]:0     You1     are2       a3    nice
dtype: object# 易错点,ix包含区间为 []In [91]: s.ix[0:0]Out[91]:0    You
dtype: objectIn [92]: s.ix[0:1]Out[92]:0    You1    are
dtype: object# ix索引不存在indexIn [95]: s.ix[400]
KeyError: 400# 按照从0开始的索引In [95]: s.iloc[0]Out[95]: 'You'In [96]: s.iloc[1]Out[96]: 'are'In [97]: s.iloc[100]
IndexError: single positional indexer is out-of-boundsIn [98]: s = pd.Series(['You','are','a','nice','girl'], index=list('abcde'))In [99]: sOut[99]:
a     You
b     are
c       a
d    nice
e    girl
dtype: objectIn [100]: s.iloc[0]Out[100]: 'You'In [101]: s.iloc[1]Out[101]: 'are'# 按照 label 索引In [103]: s.loc['a']Out[103]: 'You'In [104]: s.loc['b']Out[104]: 'are'In [105]: s.loc[['b','a']]Out[105]:
b    are
a    You
dtype: object# loc切片索引In [106]: s.loc['a':'c']Out[106]:
a    You
b    are
c      a
dtype: objectIn [108]: s.indexOut[108]: Index(['a', 'b', 'c', 'd', 'e'], dtype='object')

pandas DataFrame 索引

In [114]: import pandas as pdIn [115]: df = pd.DataFrame({'open':[1,2,3],'high':[4,5,6],'low':[6,3,1]}, index=pd.period_range('30/12/2017',perio
     ...: ds=3,freq='H'))In [116]: dfOut[116]:
                  high  low  open2017-12-30 00:00     4    6     12017-12-30 01:00     5    3     22017-12-30 02:00     6    1     3# 按列索引In [117]: df['high']Out[117]:2017-12-30 00:00    42017-12-30 01:00    52017-12-30 02:00    6Freq: H, Name: high, dtype: int64In [118]: df.highOut[118]:2017-12-30 00:00    42017-12-30 01:00    52017-12-30 02:00    6Freq: H, Name: high, dtype: int64In [120]: df[['high','open']]Out[120]:
                  high  open2017-12-30 00:00     4     12017-12-30 01:00     5     22017-12-30 02:00     6     3In [122]: df.ix[:]
D:\CodeTool\Python\python36\Scripts\ipython:1: DeprecationWarning:
.ix is deprecated. Please use
.loc for label based indexing or.iloc for positional indexingIn [123]: df.iloc[0:0]Out[123]:Empty DataFrame
Columns: [high, low, open]Index: []In [124]: df.ix[0:0]Out[124]:Empty DataFrame
Columns: [high, low, open]Index: []

# 按照 label 索引In [127]: df.indexOut[127]: PeriodIndex(['2017-12-30 00:00', '2017-12-30 01:00', '2017-12-30 02:00'], dtype='period[H]', freq='H')In [128]: df.loc['2017-12-30 00:00']Out[128]:
high    4low     6open    1Name: 2017-12-30 00:00, dtype: int64

# 检查参数In [155]: df.loc['2017-12-30 00:00:11']Out[155]:
high    4low     6open    1Name: 2017-12-30 00:00, dtype: int64In [156]: df.loc['2017-12-30 00:00:66']
KeyError: 'the label [2017-12-30 00:00:66] is not in the [index]'

填坑

In [158]: df = pd.DataFrame({'a':[1,2,3],'b':[4,5,6]}, index=[2,3,4])In [159]: dfOut[159]:
   a  b2  1  43  2  54  3  6# iloc 取第一行正确用法In [160]: df.iloc[0]Out[160]:
a    1b    4Name: 2, dtype: int64

# loc 正确用法In [165]: df.loc[[2,3]]Out[165]:
   a  b2  1  43  2  5# 注意此处 index 是什么类型In [167]: df.loc['2']
KeyError: 'the label [2] is not in the [index]'# 索引 Int64IndexOut[172]: Int64Index([2, 3, 4], dtype='int64')

# 索引为字符串In [168]: df = pd.DataFrame({'a':[1,2,3],'b':[4,5,6]}, index=list('234'))In [169]: dfOut[169]:
   a  b2  1  43  2  54  3  6In [170]: df.indexOut[170]: Index(['2', '3', '4'], dtype='object')

# 此处没有报错,千万注意 index 类型In [176]: df.loc['2']Out[176]:
a    1b    4Name: 2, dtype: int64

# ix 是一个功能强大的函数,但是争议却很大,往往是错误之源
# 咦,怎么输出与预想不一致!In [177]: df.ix[2]
D:\CodeTool\Python\Python36\Scripts\ipython:1: DeprecationWarning:
.ix is deprecated. Please use
.loc for label based indexing or.iloc for positional indexing

See the documentation here:
Http://pandas.pydata.org/pandas-docs/stable/indexing.html#ix-indexer-is-deprecatedOut[177]:
a    3b    6Name: 4, dtype: int64

# 注意开闭区间In [180]: df.loc['2':'3']Out[180]:
   a  b2  1  43  2  5

总结

pandas中ix是错误之源,大型项目大量使用它时,往往造成不可预料的后果。0.20.x版本也标记为抛弃该函数,二义性 和 []区间,违背 “Explicit is better than implicit.” 原则。建议使用意义明确的 iloc和loc 函数。

当使用字符串时切片时是 []区间 ,一般是 [)区间

当在numpy.ndarry、list、tuple、pandas.Series、pandas.DataFrame 混合使用时,采用变量进行索引或者切割,取值或赋值时,别太自信了,千万小心错误,需要大量的测试。

我在工程中使用matlab的矩阵和python混合使用以上对象,出现最多就是shape不对应,index,columns 错误。

最好不要混用不同数据结构,容易出错,更增加转化的性能开销

到此这篇关于Python基础知识之索引与切片的文章就介绍到这了,更多相关python索引与切片内容请搜索以前的文章或继续浏览下面的相关文章希望大家以后多多支持!

相关文章