Python dictionary.values() to numpy array

Question

Say I have a dict using python 3:

d = {'w1': np.array([12,11,23]), 'w2':np.array([1,2,3])}

and some variable with an integer say

a=12

How do I turn the dict values and integer into one numpy array so that it looks like this?

array([12, array([12, 11, 23]), array([1, 2, 3])])

I've tried:

np.array([a,(list(d.values()))],dtype=object)

and I get:

array([12, list([array([12, 11, 23]), array([1, 2, 3])])], dtype=object)

but I don't want the list in the second index, I want it to be "unpacked".

Are you sure you need a np.array of objects? Why not just work with a list e.g. [12] + list(d.values()) — rafaelc
– rafaelc, Commented Jul 29, 2018 at 4:18
Yeah, you don't really get much benefit out of an array of objects, especially a small one. You can't vectorize any operations for performance, or even get much convenience benefit out of it. — abarnert
– abarnert, Commented Jul 29, 2018 at 4:20
In case anyone was still looking: np.array(list(d.values())).flatten() — Shivam Chaudhary
– Shivam Chaudhary, Commented Sep 4 at 19:31

abarnert · Accepted Answer · 2018-07-29 04:15:10Z

1

If you don't want to put the arrays in an inner list, just don't put them in a list:

>>> np.array([a, *d.values()], dtype=object)
array([12, array([12, 11, 23]), array([1, 2, 3])], dtype=object)

Or, if you're on an older Python and can't unpack into a list display, create the list, but add it to another one to get the same flat list:

>>> np.array([a] + list(d.values()), dtype=object)
array([12, array([12, 11, 23]), array([1, 2, 3])], dtype=object)

answered Jul 29, 2018 at 4:15

abarnert

368k54 gold badges626 silver badges691 bronze badges

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

1 Answer 1