How to extract consecutive elements from an array containing NaN

Question

Consider following numpy array

x = np.array([1, 2, np.nan, np.nan, 3, 4, 5, np.nan])

I want to extract all non-NaN consecutive elements in x and the expected outputs is the list

y = [[1, 2],[3, 4, 5]]

Is their any method that is both elegant and faster than simple for loop ?

Answer 1

using itertools.groupby

from itertools import groupby

result = [list(map(int,g)) for k,g in groupby(x, np.isnan) if not k]
print (result)
#[[1, 2], [3, 4, 5]]

Answer 2

You can use np.split:

np.split(x, np.where(np.diff(np.isnan(x), prepend=True))[0])[1::2]
#[array([1., 2.]), array([3., 4., 5.])]