Ways to filter specific value of array in numpy

Question

I haven't use numpy to his full potential yet. I have this pretty huge 3d array (700000 x 32 x 32). All of the values are int between 1 and 4. I want to be able to filter the array so that the I get a same shape array, but only with 1 for values of 4 and 0 for everything else.

For example,

[[4, 2, 2, 3],      [[1, 0, 0, 0],
 [1, 4, 4, 2],       [0, 1, 1, 0],
 [2, 4, 1, 3],   ->  [0, 1, 0, 0],
 [2, 3, 2, 4]]       [0, 0, 0, 1]]

It works with np.where(array==4). I get a huge 3d array that I can reshape, but would there more numpy way to do it ? Thanks.

Just `(array==4).astype(int)`? – Quang Hoang Jun 10 '20 at 13:31 — Quang Hoang, Jun 10 '20 at 13:31
Yes simple as that. – Mathieu Châteauvert Jun 10 '20 at 13:41 — Mathieu Châteauvert, Jun 10 '20 at 13:41

score 2 · Accepted Answer · answered Jun 10 '20 at 13:32

arr = np.array([[4, 2, 2, 3],      
                [1, 4, 4, 2],       
                [2, 4, 1, 3],  
                [2, 3, 2, 4]])

In[58]: (arr==4).astype(int)

Out[58]: array([[1, 0, 0, 0],
                [0, 1, 1, 0],
                [0, 1, 0, 0],
                [0, 0, 0, 1]])

score 1 · Answer 2 · answered Jun 12 '20 at 12:04

1

You can also get it with np.where without having to reshape it with the next expression:

arr = np.array([[4, 2, 2, 3],      
                [1, 4, 4, 2],       
                [2, 4, 1, 3],  
                [2, 3, 2, 4]])

np.where(arr == 4, 1, 0)

Output:

array([[1, 0, 0, 0],
       [0, 1, 1, 0],
       [0, 1, 0, 0],
       [0, 0, 0, 1]])

answered Jun 12 '20 at 12:04

IMB

519
4
19

1

This is also good and fast, but the accepted answer is like 5% faster – Mathieu Châteauvert Jun 12 '20 at 12:51

score 0 · Answer 3 · answered Jun 10 '20 at 13:35

0

import numpy as np
matrix = np.zeros((5,5))
matrix[0]=[1,-2,3,4,-5]
matrix[2]=[0,1,4,0,4]
matrix=np.array([[1 if value==4 else 0 for value in row ] for row in matrix])

The above snippet uses nested list comprehension & than converts list back to np array

answered Jun 10 '20 at 13:35

Mehul Gupta

1,829
3
17
33

Yes this gets the job done well for a 2d or 3d matrix, but the use of for loops take a long time to compute. My array is pretty big. The other answer is better in my case. – Mathieu Châteauvert Jun 10 '20 at 13:40

Ways to filter specific value of array in numpy

3 Answers3