使用vstack在numpy中堆叠数组

Question

array1.shape gives (180, ) array2.shape gives (180, 1)

What's the difference between these two? And because of this difference I'm unable to stack them using

np.vstack((array2, array1))

What changes should I make to array1 shape so that I can stack them up?

Answer 1

Let's define some arrays:

>>> x = np.zeros((4, 1))
>>> y = np.zeros((4))

As is, these arrays fail to stack:

>>> np.vstack((x, y))
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib/python3/dist-packages/numpy/core/shape_base.py", line 230, in vstack
    return _nx.concatenate([atleast_2d(_m) for _m in tup], 0)
ValueError: all the input array dimensions except for the concatenation axis must match exactly

However, with a simple change, they will stack:

>>> np.vstack((x, y[:, None]))
array([[ 0.],
       [ 0.],
       [ 0.],
       [ 0.],
       [ 0.],
       [ 0.],
       [ 0.],
       [ 0.]])

Alternatively:

>>> np.vstack((x[:, 0], y))
array([[ 0.,  0.,  0.,  0.],
       [ 0.,  0.,  0.,  0.]])

Answer 2

In [81]: x1=np.ones((10,)); x2=np.ones((10,1))

One array is 1d, the other 2d. vertical stack requires 2 dimensions, vertical and horizontal. So np.vstack passes each input through np.atleast_2d :

In [82]: np.atleast_2d(x1).shape
Out[82]: (1, 10)

But now we have a (1,10) array and a (10,1) - they can't be joined in either axis.

But if we reshape x1 so it is (10,1) , then we can join it with x2 in either direction:

In [83]: np.concatenate((x1[:,None],x2), axis=0).shape
Out[83]: (20, 1)
In [84]: np.concatenate((x1[:,None],x2), axis=1).shape
Out[84]: (10, 2)

Print out the two arrays:

In [86]: x1
Out[86]: array([ 1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.])
In [87]: x2
Out[87]: 
array([[ 1.],
       [ 1.],
       [ 1.],
       [ 1.],
       [ 1.],
       [ 1.],
       [ 1.],
       [ 1.],
       [ 1.],
       [ 1.]])

How can you concatenate those two shapes without some sort of adjustment?