length of numpy.unique on a 2D list of shape (1,*)

Question

When calling numpy.unique on a list of shape (1, X) , for instance a = [[1,2]] , len(numpy.unique(a)) returns X rather than, as i'd have expected, 1. Generally speaking, numpy.unique seems to treat a as 1D array..

In contrast to this, len(numpy.array(a)) does return 1.

What is the motivation behind this?

EDIT: python version 3.7.3

Answer 1

The answer you get is because the array is flattened. And the length of [1,2] is indeed 2. I believe the reason this is done is just to give the maximum amount of flexibility to the user.

What you were looking for can be achieved by doing this instead:

len(np.unique(a, axis=0))