For example, there is a 4*2 matrix:
[[1, 2], [3, 4], [5, 6],[7,8]]
I want to reshape it to 2*2 matrix and the value of it is the mean value of the original matrix, so the result will be:
[[2,3],[6,7]]
Is there any effective way to do it? Thanks
You can reshape the array to 3d and then take mean along the second axis:
a = np.array([[1, 2], [3, 4], [5, 6],[7,8]])
step = 2
a.reshape(-1, step, a.shape[-1]).mean(1)
#array([[ 2., 3.],
# [ 6., 7.]])
Or use np.add.reduceat to add every two rows and then divide by 2:
step = 2
np.add.reduceat(a, np.arange(0,len(a),step))/step
#array([[ 2., 3.],
# [ 6., 7.]])
