count overlap between two numpy arrays

Asked May 12 '17 at 15:31

Active May 12 '17 at 15:46

Viewed 168 times

Let's say I have two arrays a and b:

a = np.arange(0,100).reshape(10,2,5)
b = np.arange(50,500).reshape(45,2,5)

I need to know how many of the 10 arrays in a arrays are in b. The following function does the job but it's not efficient enough:

def measure_overlap(array_1, array_2):
    co = 0
    for i in range(0, len(array_1)):
        if array_1[i] in array_2:
            co = co + 1
    return co

measure_overlap(a, b)
5

edited May 12 '17 at 15:46

asked May 12 '17 at 15:31

Codutie

1,055
13
25

Do you mean "10 2 dimensional arrays"? – Mad Physicist May 12 '17 at 15:35
Thank's for that.. actually it doesn't matter which shape it is, I've just removed it.. – Codutie May 12 '17 at 15:36
No problem. Fair enough. – Mad Physicist May 12 '17 at 15:37
1

Looks like I found the duplicate while working out a solution. – Mad Physicist May 12 '17 at 15:41
no it's not the correct answer. I've edited the question as a and b can also be of different shapes.. can you reopen the question? – Codutie May 12 '17 at 15:48
Sure it is. Imagine running it on `a = a.reshape(10, -1)` and it'll be the same. – Mad Physicist May 12 '17 at 15:49
1

Basically turn each of the sub-arrays into 1-D arrays containing all the data you are comparing and it becomes identical. That's what the -1 argument to `reshape` does. A more general phrasing is probably `a = a.reshape(a.shape[0], -1)` – Mad Physicist May 12 '17 at 15:51
great, it does the job very efficient; thanks! – Codutie May 12 '17 at 15:58

count overlap between two numpy arrays

0 Answers0