numpy delete rows not in list

Question

Here is my code to delete given row numbers

a = np.arange(12).reshape(6, 2)
b = [1, 4]
a: [[ 0  1]
 [ 2  3]
 [ 4  5]
 [ 6  7]
 [ 8  9]
 [10 11]]

then i delete the lines in b :

np.delete(a, b, 0)

output : 
array([[ 0,  1],
       [ 4,  5],
       [ 6,  7],
       [10, 11]])

I want to use this logic to only keep the rows in b, and do the reverse operation without any loop.
so expected output will be :

output: [
 [ 2,  3]
 [ 8,  9]
 ]

Why do you want to delete the rows? why not use a[b] to accses the part you need? You can get the desired result by a=a[b] — Erysol
– Erysol, Commented Feb 22, 2022 at 13:55
np.delete internally figures out which rows you want to keep, and returns those. So keeping the the b rows is actually simpler - conceptually and computationally. — hpaulj
– hpaulj, Commented Feb 22, 2022 at 16:55

mozway · Accepted Answer · 2022-02-22 14:04:44Z

0

IIUC, splitting the array a using the indices in b:

a2 = a[b]
a1 = np.delete(a,b,0)

output:

# a1
array([[ 0,  1],
       [ 4,  5],
       [ 6,  7],
       [10, 11]])

# a2
array([[2, 3],
       [8, 9]])

restoring the original array from a1, a2, and b:

idx = np.arange(a1.shape[0]+a2.shape[0])
idx = np.r_[np.delete(idx,b,0),idx[b]]

new_a = np.r_[a1,a2][np.argsort(idx)]

output:

array([[ 0,  1],
       [ 2,  3],
       [ 4,  5],
       [ 6,  7],
       [ 8,  9],
       [10, 11]])

answered Feb 22, 2022 at 14:04

mozway

267k13 gold badges56 silver badges106 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

user15210613 · Accepted Answer · 2022-02-23 13:56:25Z

0

The answer is just to keep the wanted rows in b by doing:

x = a[b,:]

Don't delete unwanted rows but just select the needed one

answered Feb 23, 2022 at 13:56

user15210613