ndarray对象的属性¶

In [3]:

#为了确保大家都能生成一样的数组, 我们先设置随机数种子
np.random.seed(123)
x1 = np.random.randint(1,10, size=6) # 一维数组
x2 = np.random.randint(1,10, size=(3, 4)) # 二维数组
x3 = np.random.randint(1,10, size=(3, 4, 5)) # 三维数组

In [4]:

x1
x2
x3

Out[4]:

array([3, 3, 7, 2, 4, 7])

Out[4]:

array([[2, 1, 2, 1],
       [1, 4, 5, 1],
       [1, 5, 2, 8]])

Out[4]:

array([[[4, 3, 5, 8, 3],
        [5, 9, 1, 8, 4],
        [5, 7, 2, 6, 7],
        [3, 2, 9, 4, 6]],

       [[1, 3, 7, 3, 5],
        [5, 7, 4, 1, 7],
        [5, 8, 7, 8, 2],
        [6, 8, 3, 5, 9]],

       [[2, 3, 2, 2, 4],
        [6, 1, 9, 2, 7],
        [4, 4, 6, 8, 3],
        [4, 4, 4, 9, 7]]])

数组的形状ndarray.shape¶

In [5]:

#查看数据的形状
x1.shape
x2.shape
x3.shape

Out[5]:

(6,)

Out[5]:

(3, 4)

Out[5]:

(3, 4, 5)

数组的维度ndarray.ndim¶

In [6]:

x1.ndim
x2.ndim
x3.ndim

Out[6]:

查看数组中的元素个数ndarray.size¶

In [7]:

x1.size
x2.size
x3.size

Out[7]:

查看数组中的元素的数据类型ndarray.dtype¶

In [8]:

x1.dtype
x2.dtype
x3.dtype

Out[8]:

dtype('int32')

Out[8]:

dtype('int32')

Out[8]:

dtype('int32')

查看数组中每个元素所占的内存字节数ndarray.itemsize¶

In [9]:

x1.itemsize
x2.itemsize
x3.itemsize

Out[9]:

In [10]:

# 数组的 dtype 为 int8（一个字节）
x = np.array([1,2,3,4,5], dtype = np.int8)
x.itemsize

Out[10]:

In [11]:

x = np.array([1,2,3,4,5], dtype = np.float64)
x.itemsize

Out[11]:

ndarray对象的方法一维度变换¶

In [12]:

m = np.random.randint(1, 100,(2, 3, 4))
m

Out[12]:

array([[[10, 88, 15, 84],
        [71, 13, 55, 28],
        [39, 18, 62, 75]],

       [[66, 48, 17,  6],
        [87, 47, 16, 60],
        [41, 26, 46, 50]]])

改变数组的形状ndarray.reshape（不改变原数组）¶

In [17]:

m.reshape(3,8)      #不改变原数组，会返回新的数组

Out[17]:

array([[ 1, 36, 30,  2, 84, 69, 31,  8],
       [94, 61, 66, 77, 68, 45, 52,  8],
       [89, 71, 14, 29, 64, 85, 37, 97]])

In [18]:

m.shape

Out[18]:

(2, 3, 4)

In [19]:

m.reshape(3,7)     #注意这点，要保证reshape前后数组中的元素是一样的

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-19-e3effb8f525e> in <module>
----> 1 m.reshape(3,7)     #注意这点，要保证reshape前后数组中的元素是一样的

ValueError: cannot reshape array of size 24 into shape (3,7)

改变数组的形状ndarray.resize(改变原数组)¶

In [20]:

m = np.random.randint(1, 100,(2, 3, 4))
m

Out[20]:

array([[[41, 89, 64, 59],
        [78,  9, 79,  7],
        [66, 95, 71, 41]],

       [[75, 77, 77, 26],
        [ 8, 14, 45,  2],
        [42, 79, 57, 88]]])

In [21]:

m.resize(4,6) #改变原数组

In [22]:

Out[22]:

array([[41, 89, 64, 59, 78,  9],
       [79,  7, 66, 95, 71, 41],
       [75, 77, 77, 26,  8, 14],
       [45,  2, 42, 79, 57, 88]])

In [23]:

m.shape

Out[23]:

(4, 6)

In [27]:

m.shape

Out[27]:

(2, 3, 4)

返回一个一维数组¶

ndarray.flatten¶

In [39]:

m = np.random.randint(1, 100,(2, 3, 4))
m

Out[39]:

array([[[ 6, 82, 36, 78],
        [26, 98, 94, 71],
        [59, 30, 82, 22]],

       [[17, 31, 56, 81],
        [31, 13, 91, 84],
        [17, 93, 39, 99]]])

In [40]:

n = m.flatten()  #不改变原数组
n 

Out[40]:

array([ 6, 82, 36, 78, 26, 98, 94, 71, 59, 30, 82, 22, 17, 31, 56, 81, 31,
       13, 91, 84, 17, 93, 39, 99])

In [ ]:

np.ravel()¶

In [42]:

Out[42]:

array([[[ 6, 82, 36, 78],
        [26, 98, 94, 71],
        [59, 30, 82, 22]],

       [[17, 31, 56, 81],
        [31, 13, 91, 84],
        [17, 93, 39, 99]]])

In [43]:

m.ravel()

Out[43]:

array([ 6, 82, 36, 78, 26, 98, 94, 71, 59, 30, 82, 22, 17, 31, 56, 81, 31,
       13, 91, 84, 17, 93, 39, 99])

In [ ]:

In [38]:

Out[38]:

array([[[36, 59, 39,  2],
        [53, 18, 46, 23],
        [54, 90, 75,  1]],

       [[36, 84, 95, 72],
        [40, 88, 84, 79],
        [30, 20, 38,  2]]])

通过reshape改变形状¶

In [46]:

m.reshape(1,-1)

Out[46]:

array([[ 6, 82, 36, 78, 26, 98, 94, 71, 59, 30, 82, 22, 17, 31, 56, 81,
        31, 13, 91, 84, 17, 93, 39, 99]])

In [47]:

m.reshape(-1,1)

Out[47]:

array([[ 6],
       [82],
       [36],
       [78],
       [26],
       [98],
       [94],
       [71],
       [59],
       [30],
       [82],
       [22],
       [17],
       [31],
       [56],
       [81],
       [31],
       [13],
       [91],
       [84],
       [17],
       [93],
       [39],
       [99]])

这种改变的就是原数组的形状而已

In [ ]:

ndarray对象的方法—数组转列表ndarray.tolist¶

有时候我们需要把把数组转化为列表来处理数据，因为列表有很多方法

之前我们有学过ndarray.astype()用来修改数组元素的数据类型

也知道怎么把列表元组变成数组，反过来我再学一个怎么把ndarray变成列表

In [58]:

a  = np.array([1, 2, 3, 4, 5, 6, 7, 8])   #列表变数组
a

Out[58]:

array([1, 2, 3, 4, 5, 6, 7, 8])

In [57]:

b = a.tolist()  #不会改变原数组，重新返回一个新生成的列表
b

Out[57]:

[1, 2, 3, 4, 5, 6, 7, 8]

In [53]:

Out[53]:

array([1, 2, 3, 4, 5, 6, 7, 8])

ndarray的索引与切片¶

和字符串列表元祖基本上是一样的
[开始位置, 终止位置，步长和方向]

一维ndarray的索引与切片¶

In [13]:

a = np.arange(5,16)
a

Out[13]:

array([ 5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15])

索引单个元素

In [14]:

a[1]

Out[14]:

In [15]:

a[-1]

Out[15]:

In [16]:

a[-2]

Out[16]:

多个元素的连续索引

In [17]:

a[0:3]

Out[17]:

array([5, 6, 7])

In [18]:

a[3:8]

Out[18]:

array([ 8,  9, 10, 11, 12])

In [19]:

a[5:]

Out[19]:

array([10, 11, 12, 13, 14, 15])

In [20]:

a[5:-1]

Out[20]:

array([10, 11, 12, 13, 14])

间隔索引

In [21]:

Out[21]:

array([ 5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15])

In [22]:

a[0:6:2]

Out[22]:

array([5, 7, 9])

In [23]:

a[0::4]

Out[23]:

array([ 5,  9, 13])

In [24]:

a[::-2]

Out[24]:

array([15, 13, 11,  9,  7,  5])

多维ndarray的索引与切片¶

In [25]:

a = np.array([[1, 2, 3], [4, 5, 6,], [7, 8, 9]])
a

Out[25]:

array([[1, 2, 3],
       [4, 5, 6],
       [7, 8, 9]])

In [26]:

a[0]

Out[26]:

array([1, 2, 3])

In [27]:

a[0:2]

Out[27]:

array([[1, 2, 3],
       [4, 5, 6]])

In [28]:

a[0][1]

Out[28]:

In [29]:

a[2][:]

Out[29]:

array([7, 8, 9])

In [30]:

Out[30]:

array([[1, 2, 3],
       [4, 5, 6],
       [7, 8, 9]])

In [31]:

a[0:2,0]   #[行,列]

Out[31]:

array([1, 4])

In [32]:

a[:,:]

Out[32]:

array([[1, 2, 3],
       [4, 5, 6],
       [7, 8, 9]])

In [33]:

a[:,1]

Out[33]:

array([2, 5, 8])

In [34]:

a[1]==a[1,:]

Out[34]:

array([ True,  True,  True])

In [ ]:

布尔索引¶

当结果对象是布尔运算（例如比较运算符）的结果时，将使用此类型的高级索引。

In [35]:

x = np.arange(12).reshape(4,3)
x

Out[35]:

array([[ 0,  1,  2],
       [ 3,  4,  5],
       [ 6,  7,  8],
       [ 9, 10, 11]])

In [36]:

x>5

Out[36]:

array([[False, False, False],
       [False, False, False],
       [ True,  True,  True],
       [ True,  True,  True]])

In [37]:

x[x>5]

Out[37]:

array([ 6,  7,  8,  9, 10, 11])

In [38]:

a = np.array([np.nan,1,2,np.nan,3,4,5])
a

Out[38]:

array([nan,  1.,  2., nan,  3.,  4.,  5.])

In [39]:

np.isnan(a)

Out[39]:

array([ True, False, False,  True, False, False, False])

In [40]:

a[np.isnan(a)]

Out[40]:

array([nan, nan])

In [ ]:

Table of Contents

ndarray对象的属性¶

数组的形状ndarray.shape¶

数组的维度ndarray.ndim¶

查看数组中的元素个数ndarray.size¶

查看数组中的元素的数据类型ndarray.dtype¶

查看数组中每个元素所占的内存字节数ndarray.itemsize¶

ndarray对象的方法一维度变换¶

改变数组的形状ndarray.reshape（不改变原数组）¶

改变数组的形状ndarray.resize(改变原数组)¶

返回一个一维数组¶

ndarray.flatten¶

np.ravel()¶

通过reshape改变形状¶

ndarray对象的方法—数组转列表ndarray.tolist¶

ndarray的索引与切片¶

一维ndarray的索引与切片¶

多维ndarray的索引与切片¶

布尔索引¶