vectorize array from dictionary with tuple as key

Question

This is a very simple task, but somehow, I can't seem to figure it out.

Let's say I have a dictionary

dictr = {(0,1,2): 0, (5,4,2):1}

And an array:

A = [[0, 1, 2], [5, 4, 2]]

I am looking to get an array b = [0, 1]

I thought we could do it by:

B = np.vectorize(dictr.get)(A)

However, this is not working. Anyone knows why?

np.vectorize passes scalar elements of A to get, not pairs or tuples. Sometimes vectorize is convenient, but it does not help with speed. And often it is hard to use correctly. — hpaulj
– hpaulj, Commented Apr 8, 2020 at 18:07

tstanisl · Accepted Answer · 2020-04-08 09:40:42Z

1

The problem is that dictr.get is called for elements of A, not for A's rows. It calls dictr.get(A[0,0]), next dictr.get(A[0,1]), and so on. Each of elements of A is a scalar not present in the dictionary thus np.vectorize fallbacks to None. That is why you get 2x3 array of None.

answered Apr 8, 2020 at 9:40

tstanisl

14.3k3 gold badges31 silver badges45 bronze badges

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

Comments

Hoàng Tùng · Accepted Answer · 2020-04-08 09:32:09Z

0

It is because your dictionary keys are tuple, meanwhile the items in A are list. You should convert all the element inside A to tuple so the key may match

A python-ic way (without numpy) might seems like this (considering every item in A matches a key in dictr)

b = [dictr[tuple(item)] for item in A]

answered Apr 8, 2020 at 9:32

Hoàng Tùng

1301 gold badge2 silver badges10 bronze badges

1 Comment

Hoàng Tùng Over a year ago

Remember that this code will produce error if there is an element in A which doesn't match with a key in dictr, so make sure to check it if you are not sure about the condition ;)

Collectives™ on Stack Overflow

vectorize array from dictionary with tuple as key

2 Answers 2

Comments

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related