Let's say I have an numpy array a = [1,2,3,4,5] and another numpy array b = [6,7,8,9,10]. How do I concatenate the arrays and create a numpy array of shape (5,2). I tried both np.row_stack and np.vstack (which do the same thing), but that only results in an np array of shape (2,5). How would I avoid this and get the desired concatenation?
Than you!
np.vstack is the way to go, you just need an added transposition step afterwards.
c = np.vstack((a, b)).T
print(c)
array([[ 1, 6],
[ 2, 7],
[ 3, 8],
[ 4, 9],
[ 5, 10]])
print(c.shape)
(5, 2)
As djk47463 mentioned, you can also use the np.array constructor (will delete if they decide to post):
c = np.array([a,b]).T
print(c)
array([[ 1, 6],
[ 2, 7],
[ 3, 8],
[ 4, 9],
[ 5, 10]])
Newish np.stack gives more control over the new axis:
In [37]: a = [1,2,3,4,5]
In [38]: b = [6,7,8,9,10]
In [39]: np.stack((a,b), axis=1)
Out[39]:
array([[ 1, 6],
[ 2, 7],
[ 3, 8],
[ 4, 9],
[ 5, 10]])
In [40]: _.shape
Out[40]: (5, 2)
With the default axis=0, it behaves like np.array, producing a (2,5) array.
vstack docs notes:
This function continues to be supported for backward compatibility, but you should prefer
np.concatenateornp.stack. Thenp.stackfunction was added in NumPy 1.10.
I think that's overstating the case, but still, stack is one of my favorite new functions. I also recommend looking at the Python code for functions like this. Most end up using concatenate after fiddling with the dimensions of the inputs.
np.vstack((a, b)).Tnp.array([a,b]).T