I have the following numpy array:
X = [[1],
[2],
[3],
[4]]
Y = [[5],
[6],
[7],
[8]]
Z = [[9],
[10],
[11],
[12]]
I would like to get the following output:
H = [[1,5,9],
[2,6,10],
[3,7,11]
[4,8,12]]
Is there a way get this result using numpy.reshape?
You can use np.column_stack -
np.column_stack((X,Y,Z))
Or np.concatenate along axis=1 -
np.concatenate((X,Y,Z),axis=1)
Or np.hstack -
np.hstack((X,Y,Z))
Or np.stack along axis=0 and then do multi-dim transpose -
np.stack((X,Y,Z),axis=0).T
Reshape applies on an array, not to stack or concatenate arrays together. So, reshape alone doesn't make sense here.
One could argue using np.reshape to give us the desired output, like so -
np.reshape((X,Y,Z),(3,4)).T
But, under the hoods its doing the stacking operation, which AFAIK is something to convert to array with np.asarray -
In [453]: np.asarray((X,Y,Z))
Out[453]:
array([[[ 1],
[ 2],
[ 3],
[ 4]],
[[ 5],
[ 6],
[ 7],
[ 8]],
[[ 9],
[10],
[11],
[12]]])
We just need to use multi-dim transpose on it, to give us a 3D array version of the expected output -
In [454]: np.asarray((X,Y,Z)).T
Out[454]:
array([[[ 1, 5, 9],
[ 2, 6, 10],
[ 3, 7, 11],
[ 4, 8, 12]]])
np.concatenate is overkill. They either do unneeded shape tests, or create a 3d array (stack and asarray variants).And don't forget np.c_ (I don't see the need for np.reshape):
np.c_[X,Y,Z]
# array([[ 1, 5, 9],
# [ 2, 6, 10],
# [ 3, 7, 11],
# [ 4, 8, 12]])
How about this (faster) solution?
In [16]: np.array([x.squeeze(), y.squeeze(), z.squeeze()]).T
Out[16]:
array([[ 1, 5, 9],
[ 2, 6, 10],
[ 3, 7, 11],
[ 4, 8, 12]])
Efficiency (descending order)
# proposed (faster) solution
In [17]: %timeit np.array([x.squeeze(), y.squeeze(), z.squeeze()]).T
The slowest run took 7.40 times longer than the fastest. This could mean that an intermediate result is being cached.
100000 loops, best of 3: 7.36 µs per loop
# Other solutions
In [18]: %timeit np.column_stack((x, y, z))
The slowest run took 5.18 times longer than the fastest. This could mean that an intermediate result is being cached.
100000 loops, best of 3: 9.18 µs per loop
In [19]: %timeit np.hstack((x, y, z))
The slowest run took 4.49 times longer than the fastest. This could mean that an intermediate result is being cached.
100000 loops, best of 3: 16 µs per loop
In [20]: %timeit np.reshape((x,y,z),(3,4)).T
10000 loops, best of 3: 21.6 µs per loop
In [20]: %timeit np.c_[x, y, z]
10000 loops, best of 3: 55.9 µs per loop
