Pythonic way to cut specific elements from numpy array

Question

I have a Python list (numpy array) and another list which contains the indices for the location of values from the first array which I want to keep.

Is there a Pythonic way to do this? I know numpy.delete, but I want to keep the elements and not delete them.

What have you tried? What's your existing solution that you think that may not be "Pythonic" enough? — woozyking
– woozyking, Commented Sep 15, 2018 at 6:21
@woozyking option 1: iterating over the main array and checking if that index is there in the second array or not (we can convert the second list to a map for O(1) lookup). option 2: make a copy of first array and then deleting using numpy.delete and then taking the difference between the original list and new (smaller) list - but this will not retain the original order. — Karan_Saxena
– Karan_Saxena, Commented Sep 15, 2018 at 6:26
That's wonderful to have more than one option. However it'd be even better for you to actually demonstrate them as (sample) code so others can help you more effectively. — woozyking
– woozyking, Commented Sep 15, 2018 at 6:30
@woozyking I'm not sure if a (sample) code will help. My question is not about how to efficiently implement what I wrote, but if there's a fundamentally different way to achieve this (perhaps in a single line) which I am missing. — Karan_Saxena
– Karan_Saxena, Commented Sep 15, 2018 at 6:34
I'm not sure what your starting conditions are. Do you want to act on a Python list or a numpy.array? Obviously, it cannot be both at the same time. — applesoup
– applesoup, Commented Sep 15, 2018 at 11:20

applesoup · Accepted Answer · 2018-09-15 11:18:39Z

2

The most pythonic way is probably also the most straightforward one:

a = np.array([2,5,6,3,6,3,45,6])
b = [0,3,4,7] # indices that you need to keep
c = a[b]

or, if you do not need a any longer:

a = a[b]

answered Sep 15, 2018 at 11:18

applesoup

4878 silver badges10 bronze badges

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

Comments

Srikar · Accepted Answer · 2018-09-15 06:32:30Z

1

You can create a new list with the values that you want to keep.

a = np.array([2,5,6,3,6,3,45,6])
b = [0,3,4,7] #indices that yo need to keep
c = [a[i] for i in b]

answered Sep 15, 2018 at 6:32

Srikar

711 silver badge5 bronze badges

Well, this is more cleaner than any of the approaches I mentioned above. Thanks for this!

Why not just a[b]? Much more efficient to use numpy indexing if working with numpy arrays

msi_gerva · Accepted Answer · 2018-09-15 11:14:57Z

1

Why don't you use just c=a[b] as this is the Python way to take the values from array a.

answered Sep 15, 2018 at 11:14

msi_gerva

2,0803 gold badges22 silver badges31 bronze badges