20 methods for creating ndarrary in Python

Complete example of this article: complete sample code
This article describes the basic, common ways to create ndarrary, with sample code.

First, create through Ndarray

importas np

1.1 One-dimensional arrays

= np.array([123])a

array([1, 2, 3])

1.2 Two-dimensional arrays

np.array([[1234],       [2345]])

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

1.3 Three-dimensional array

= np.array([    [        [123223],        [233233]    ],    [        [1234],        [233432]    ]])print(arr1)print(type(arr1))arr1

[[[ 1  2 32 23]  [23  3 23  3]] [[ 1  2  3  4]  [23  3  4 32]]]<class ‘numpy.ndarray‘>array([[[ 1,  2, 32, 23],        [23,  3, 23,  3]],       [[ 1,  2,  3,  4],        [23,  3,  4, 32]]])

Second, the common function of creating Ndarrary 2. The zeros () function, which specifies a dimension tuple (shape parameter), returns a total of 0 ndarrary

= np.zeros((24))arr2

array([[ 0.,  0.,  0.,  0.],       [ 0.,  0.,  0.,  0.]])

3. Ones () function, specify a dimension tuple (shape parameter), return all 1 ndarrary

= np.ones((44))arr3

array([[ 1.,  1.,  1.,  1.],       [ 1.,  1.,  1.,  1.],       [ 1.,  1.,  1.,  1.],       [ 1.,  1.,  1.,  1.]])

4. Empty () function, specifying a dimension tuple (shape parameter), returns a value (garbage value) as the initialized Ndarrary

= np.empty((22))arr4

array([[  7.89119642e-312,   4.22795269e-307],       [  9.34608432e-307,   1.11258854e-306]])

Iii. other ways to create ndarrary 5. Numpy.arange ([start], stop[, step], dtype=none) function, left open right closed

= np.arange(1101)arr5

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

6. Numpy.linspace (Start, Stop, num=50, Endpoint=true, Retstep=false, dtype=none) function, generate a arithmetic progression, left closed right

= np.linspace(045)arr6

array([ 0.,  1.,  2.,  3.,  4.])

7. Numpy.logspace (Start, Stop, num=50, Endpoint=true, base=10.0, Dtype=none) functions, base default is 10

= np.logspace(045, base=2)arr7

array([  1.,   2.,   4.,   8.,  16.])

8. Numpy.eye (N, M=none, k=0, dtype=

Return a 2-d array with ones on the diagonal and zeros elsewhere.

= np.eye(43)arr8

array([[ 1.,  0.,  0.],       [ 0.,  1.,  0.],       [ 0.,  0.,  1.],       [ 0.,  0.,  0.]])

9. Numpy.zeros_like (A, Dtype=none, order= ' K ', subok=true) function, returns an array of all 0 elements with the same shape parameter as a

Return an array of zeros with the same shape and type as a given array.
Parameter subok, whether to inherit the data type of A;
The parameter order, which specifies the stored layout of the returned results in memory, is ' K ' by default, indicating as much as possible as a.

# 返回一个与arr8形状应的全0数组= np.zeros_like(arr8)arr9

array([[ 0.,  0.,  0.],       [ 0.,  0.,  0.],       [ 0.,  0.,  0.],       [ 0.,  0.,  0.]])

Numpy.ones_like (A, Dtype=none, order= ' K ', subok=true) function,

Returns an array that is the same as the shape parameter of a and the element is all 0

# 返回一个与ar9形状应的全0数组= np.ones_like(arr9)arr10

array([[ 1.,  1.,  1.],       [ 1.,  1.,  1.],       [ 1.,  1.,  1.],       [ 1.,  1.,  1.]])

Numpy.empty_like (A, Dtype=none, order= ' K ', subok=true) functions, returns an array with the same shape and all values as garbage values (random values)

Return a new array with the same shape and type as a given array.

= np.empty_like(np.ones((5,4)))arr11

array([[  7.89102294e-312,   6.27463370e-322,   0.00000000e+000,          0.00000000e+000],       [  8.45593933e-307,   5.30276956e+180,   7.70748458e-043,          4.57487963e-071],       [  3.45618033e-086,   3.35860426e-143,   6.01433264e+175,          6.93885958e+218],       [  5.56218858e+180,   3.94356143e+180,   4.75084178e-037,          1.24689504e-047],       [  3.85156077e-057,   2.06073242e+184,   4.71530148e-143,          1.50008929e+248]])

Numpy.copy (A, order= ' K ') function, returns the same array as a

Return An array copy of the given object.

# 属于深拷贝，修改复制数组，对原数组没有影响= np.copy(arr11)arr12[00=1arr12

array([[  1.00000000e+000,   6.27463370e-322,   0.00000000e+000,          0.00000000e+000],       [  8.45593933e-307,   5.30276956e+180,   7.70748458e-043,          4.57487963e-071],       [  3.45618033e-086,   3.35860426e-143,   6.01433264e+175,          6.93885958e+218],       [  5.56218858e+180,   3.94356143e+180,   4.75084178e-037,          1.24689504e-047],       [  3.85156077e-057,   2.06073242e+184,   4.71530148e-143,          1.50008929e+248]])

Numpy.identity (n, dtype=none) function, returns an n-order unit square

Return the identity array.

= np.identity(4)arr13

array([[ 1.,  0.,  0.,  0.],       [ 0.,  1.,  0.,  0.],       [ 0.,  0.,  1.,  0.],       [ 0.,  0.,  0.,  1.]])

Numpy.fromfunction (function, shape, **kwargs) returns an array in which the value of the element is computed by the function, and the shape is limited by the form tuple
Note: The parameter shape is a tuple, and the number of elements in the shape tuple needs to match the number of parameters in the function, representing different latitudes.

= np.fromfunction(lambda x, y: x+y, shape=(33))print= np.fromfunction(lambda x, y, z: x+y+z, shape=(333))arr21

[[ 0.  1.  2.] [ 1.  2.  3.] [ 2.  3.  4.]]array([[[ 0.,  1.,  2.],        [ 1.,  2.,  3.],        [ 2.,  3.,  4.]],       [[ 1.,  2.,  3.],        [ 2.,  3.,  4.],        [ 3.,  4.,  5.]],       [[ 2.,  3.,  4.],        [ 3.,  4.,  5.],        [ 4.,  5.,  6.]]])

Numpy.mgrid functions

arr14 =  np.mgrid[ - 1 : 3 : 2 ] # means starting from 1, step 1, Take 2 numbers  print  (arr14) arr15 =  np.mgrid[-  1 : 3 : 2j] # when added ' J ', means left closed right closed, and takes 2 numbers  span class= "bu" >print  (arr15) arr16 =  np.mgrid[- 1 : 3 ] # when specifying two parameters, the function is the same as the Numpy.arange () function  span class= "bu" >print  (arr16) arr18 =  np.mgrid[0 : 5 , 0 : 5 ] # do padding on two dimensions respectively  print  (arr18)  

[-1  1][-1.  3.][-1  0  1  2][[[0 0 0 0 0]  [1 1 1 1 1]  [2 2 2 2 2]  [3 3 3 3 3]  [4 4 4 4 4]] [[0 1 2 3 4]  [0 1 2 3 4]  [0 1 2 3 4]  [0 1 2 3 4]  [0 1 2 3 4]]]

Numpy.ogrid functions

= np.ogrid[1:2]print= np.ogrid[0:50:5]print(arr20)

[1][array([[0],       [1],       [2],       [3],       [4]]), array([[0, 1, 2, 3, 4]])]

Numpy.fromstring (String, Dtype=float, Count=-1, sep= ") function to create a one-dimensional ndarrary from a string

= np.fromstring(‘1 2 3 4‘, sep=‘ ‘)arr21

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

Numpy.fromiter (iterable, Dtype, count=-1) function, returns a one-dimensional array

# 需要指定数据类型dtype= (i*forinrange(4= np.fromiter(iterable,dtype=float)arr22

array([ 0.,  1.,  4.,  9.])

Iv. other methods 19. The Numpy.fromfile (file, Dtype=float, Count=-1, sep= ") function creates ndarrary20 from a text file or binary file. numpy.loadtxt(fname, dtype=<type ‘float’>, comments=’#’, delimiter=None, converters=None,skiprows=0, usecols=None, unpack=False, ndmin=0)Created from the file.

Summarize

1. Complete example of this article: complete sample code
2. Limited capacity, welcome to the wrong communication;

Welcome to the personal public number Waltsmithml or Sina Weibo waltsmith, which offers free video tutorials on machine learning, deep learning, Hadoop, Spark, Python, Math, and more. My main direction is machine learning and deep learning. Very welcome to exchange study ha, in addition to study, but also free to help download paper or books ha ==============
??? Public number????????? Sina micro-blog??

20 methods for creating ndarrary in Python