np.newaxis

在原有得到矩阵中增加一个轴，

实际上上np.newaxis的值为无

print(np.newaxis)
#None

np.newaxis的位置决定要在哪个维度增加

import numpy as np

a = np.arange(1, 10)
a = np.reshape(a, (3, 3))
print(a[:, np.newaxis].shape)
'''
shape(3, 1, 3)
array([[1, 2, 3],
       [4, 5, 6],
       [7, 8, 9]])
'''
a = np.mean(a, axis=1)
'''
a = [2. 5. 8.]
a.shape = (3,)
'''
a = a[:, np.newaxis, np.newaxis]
'''
array([[[[[2.]]]],
       [[[[5.]]]],
       [[[[8.]]]]])
shape(3, 1, 1)
'''
a[:, np.newaxis, np.newaxis]