【python】numpy中axis和keepdims的直观解释

import numpy as np

x = np.array([[1,2,3],
              [4,5,6],
              [7,8,9]])
add = np.sum(x)
add_axis = np.sum(x, axis=1)
add_keepdims = np.sum(x, axis=1, keepdims = True)

print('add = ',add)
print('add_axis_1 = ',add_axis, 'shape = ', add_axis.shape)
print('add_axis_1_keepdims  = ',add_keepdims)

以上代码输出的结果如下所示：

>add =  45
>add_axis_1 =  [ 6 15 24]     shape =  (3,)   #注意这里的shape
>add_axis_1_keepdims  =  [[ 6]
 [15]
 [24]]

可以看出，

1. np.sum是对数组的全部元素求和
2. 令axis = 1，是对列向量进行降维操作，把列向量压缩即是对数组的行向量求和，得出的结果不具有维度，是标量
3. keepdims = 1，是保持数组维度，在这个例子中所以是对列向量进行压缩，所以，应该还是3行，正如结果所示，得到的是3*1的结果

import numpy as np

x = np.array([[1,2,3],