Find the index of the next value greater than threshold in Numpy

Question

Suppose I've got an array a,

a = np.array([50,49,47,55,68,70,65])

And I want the index of the first value that is greater than 65 (i.e.4). What is the fastest way I can do this task in Numpy (since my actual dataset is much larger)?

Linear? Your array isn't sorted, you can't hope for anything much better. — cs95, Mar 28 '19 at 09:58
@coldspeed Why close? He wanted best performance and time complexity, it's different case. Plus you can use at most N operations, not at least N as is the case with `argmax`. — Szymon Maszke, Mar 28 '19 at 10:04
@coldspeed my bad, indeed, mobile is not helpful, apologies. — Szymon Maszke, Mar 28 '19 at 10:07
You meant to ping @Divakar (btw, you can edit your comment upto 5 minutes, and you can change the user you ping). — cs95, Mar 28 '19 at 10:09

score 0 · Answer 1 · answered Mar 28 '19 at 09:57

0

argmax will find the index of first True, try this np.argmax(a>65)

answered Mar 28 '19 at 09:57

Lucien

118
1
6

score 0 · Answer 2 · answered Mar 28 '19 at 10:01

I would suggest using numba and it's decorators. Thanks to this library you can achieve better performance scores than numpy (or on par) while using plain Python.

In your case it could be:

import numba

@numba.njit
def find_greater(array, threshold: int):
    for i in range(len(array)):
        if array[i] > threshold:
            return i

This option is better in terms of time complexity, maximum O(n) (usually much less) than approaches like the one suggested in comments.

Find the index of the next value greater than threshold in Numpy

2 Answers2