I have such an array:
arr = np.array([1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16]).reshape(2,2,4)
array([[[ 1, 2, 3, 4],
[ 5, 6, 7, 8]],
[[ 9, 10, 11, 12],
[13, 14, 15, 16]]])
And I want to get to get the max for each of these sub-arrays:
arr[:,:,-1]
array([[ 4, 8],
[12, 16]])
So I would want each of these sub-arrays converted to their max value. As result:
array([8, 16)]
How would I be able to do that without iterating?
Use max in numpy:
arr[:,:,-1].max(-1)
-1 is your last dimension.
printing it:
[8, 16]
arr[:,:,-1].max(-1)? It will give you [8,16]. but if you mean something else by aggregation, could you please explain more how you get the output? Thank youarr.max(-1).max(-1) if you were looking for this.IIUC you want the output to be [8,16] which is a max of the sub-arrays arr[:,:,-1] that you have computed. Try this -
arr = np.array([1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16]).reshape(2,2,4)
np.max(arr[:,:,-1],-1)
array([ 8, 16])
