如何转置Numpy数组的例子

Numpy提供了4种方法对数组对象进行转置。它们是rollaxis(),swapaxes(),transpose(), ndarray**.T**。本文将向你展示一些如何转置Numpy数组的例子。

1.numpy.rollaxis()

1. 当输入的数组是一个多维数组时，那么你可以使用这个方法将指定的数组轴移动到指定的位置。

2. 下面是该方法的格式：

   numpy.rollaxis(arr, axis, start)
   

3. arr：传入的多维Numpy数组。

4. axis：要移动的多维数组轴的索引。其他轴的相对位置不改变。

5. start：移动轴的新索引。

6. 但是从Numpy 1.11.0版本开始，你应该使用moveaxis(array, source, destination)方法来代替rollaxis()方法。

7. 下面是例子的源代码，你可以看到注释中的详细解释。

   import numpy as np
   
   def transpose_numpy_array_rollaxis():
       
       # create the original 3 dimensional array that has 5 rows (axis 0), 2 columns(axis 1), and each element is an array that has 3 values(axis 2).
       # there are 3 axis in the array axis_0 - 5 rows, axis_1 - 2 columns, axis_2 - 3 values.
       array = np.arange(30).reshape(5,2,3)
       print('********** original array **********\r')
       print(array)
       print('\r array.ndim = ', array.ndim)
       
       
       # move the third axis (axis_2) to the second axis ( axis_1), that means swap the 2 axis.
       array_rolled_1 = np.rollaxis(array, axis = 2, start = 1)
       print('\n********** after roll the array\'s third axis to the second axis **********\r')
       # after the above swap, the array has 5 rows (axis 0), 3 columns ( axis 1), each element array has 2 values(axis 2).
       print(array_rolled_1)
       print('\r array_rolled_1.ndim = ', array_rolled_1.ndim)
       
       # move the third axis (axis_2) to the first axis ( axis_0), that means swap the 2 axis.
       array_rolled_2 = np.rollaxis(array, axis = 2, start = 0)
       print('\n********** after roll the array\'s third axis to the first axis **********\r')
       # after the above swap, the array has 3 rows (axis 0), 5 columns ( axis 1), each element array has 2 values(axis 2).
       print(array_rolled_2)
       print('\r array_rolled_2.ndim = ', array_rolled_2.ndim)
       
       
       array1 = np.arange(10).reshape(5,2)
       print(array1)
       print('\n original array = ', array1)
       # move the axis 1 to the first axis that means swap the axis 1 and axis 0.
       array1_rolled = np.moveaxis(array1, source = 1, destination = 0)
       print('\n np.moveaxis(array1, source = 1, destination = 0) = ', array1_rolled)
                                                
   
   if __name__ == '__main__':
       
       numpy_ndarray_ravel_example()
   
   

8. 下面是上述例子的源代码执行结果：

   ********** original 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]]]
   
    array.ndim =  3
   
   ********** after roll the array's third axis to the second axis **********
   
   [[[ 0  3]
     [ 1  4]
     [ 2  5]]
   
    [[ 6  9]
     [ 7 10]
     [ 8 11]]
   
    [[12 15]
     [13 16]
     [14 17]]
   
    [[18 21]
     [19 22]
     [20 23]]
   
    [[24 27]
     [25 28]
     [26 29]]]
   
    array_rolled_1.ndim =  3
   
   ********** after roll the array's third axis to the first axis **********
   
   [[[ 0  3]
     [ 6  9]
     [12 15]
     [18 21]
     [24 27]]
   
    [[ 1  4]
     [ 7 10]
     [13 16]
     [19 22]
     [25 28]]
   
    [[ 2  5]
     [ 8 11]
     [14 17]
     [20 23]
     [26 29]]]
   
    array_rolled_2.ndim =  3
   [[0 1]
    [2 3]
    [4 5]
    [6 7]
    [8 9]]
   
    original array =  [[0 1]
    [2 3]
    [4 5]
    [6 7]
    [8 9]]
   
    np.moveaxis(array1, source = 1, destination = 0) =  [[0 2 4 6 8]
    [1 3 5 7 9]]
   
   

2.numpy.swapaxes()

1. 该方法用于交换一个数组的两个轴。

2. 该方法的语法如下：

   numpy.swapaxes(arr, axis1, axis2)
   

3. 下面是一个例子：

   import numpy as np
   
   def transpose_numpy_array_swapaxes():
       
       # create the original 2 dimensional array. 
       array = np.arange(10).reshape(5,2)
       
       print('\n source array = ', array)
       
       # swap the 2 axis of the 2 dimensional array.
       array_1 = np.swapaxes(array,1,0)
       
       print('\n np.swapaxes(array,1,0) = ', array_1)
                                            
   
   if __name__ == '__main__':
       
       transpose_numpy_array_swapaxes()
   

4. 下面是例子的执行输出：

    source array =  [[0 1]
    [2 3]
    [4 5]
    [6 7]
    [8 9]]
   
    np.swapaxes(array,1,0) =  [[0 2 4 6 8]
    [1 3 5 7 9]]
   
   

3.Numpy.transpose()

1. Numpy.transpose()方法用于交换多维数组的维数。

2. 下面是Numpy.transpose()方法的语法。

   numpy.transpose(arr, axes)
   

3. arr：要操作的数组。

4. axes：可选参数，元组或整数列表，它将根据该参数对数组进行转置。

5. 下面是例子的源代码。

   import numpy as np
   
   def transpose_numpy_array_transpose():
       
       # create the original 2 dimensional array. 
       array = np.arange(10).reshape(5,2)
       
       print('\n source array = ', array)
       
       # swap the 2 axis of the 2 dimensional array.
       array_1 = np.transpose(array)
       
       print('\n np.transpose() = ', array_1)                                             
   
   if __name__ == '__main__':
       
       transpose_numpy_array_transpose()
   

6. 下面是上述例子的执行输出：

   source array =  [[0 1]
    [2 3]
    [4 5]
    [6 7]
    [8 9]]
   
    np.transpose() =  [[0 2 4 6 8]
    [1 3 5 7 9]]
   

4. ndarray.T.

1. 这个属性类似于numpy.transpose()方法。

   import numpy as np
   
   def transpose_numpy_array_t():
       
       # create the original 2 dimensional array. 
       array = np.arange(10).reshape(5,2)
       
       print('\n source array = ', array)
       
       # swap the 2 axis of the 2 dimensional array.
       array_1 = array.T
       
       print('\n array.T = ', array_1)                                                 
   
   if __name__ == '__main__':
       
       transpose_numpy_array_t()
   

2. 下面是上述例子的执行输出。

   source array =  [[0 1]
    [2 3]
    [4 5]
    [6 7]
    [8 9]]
   
    array.T =  [[0 2 4 6 8]
    [1 3 5 7 9]]