Constructing a ndarray of values from another ndarray containing dictionary keys

Question

I have a ndarray containing the dictionary keys arranged in a specific order. I want to create another ndarray containing the values of the respective keys. The order have to be maintained. Obvious approach is to iterate over the array containing the keys element by element but the problem is there is no way to know the shape of the array beforehand.

Is it possible to flatten the ndarray of keys and iterate over it to generate flat ndarray of values and finally unravel it without harming the order?

 mydict = {'a':1, 'b':2, 'c':3, 'd':4, 'e':5, 'f':6}
 input_pattern = np.array([['a', 'f'], ['b', 'e'], ['c', 'd']])
 expected_output = np.array([[1, 6], [2, 5], [3, 4]])

N. B. : In the above example pattern array is 2D, it may not be the case always. Also it may not contain all keys of the dictionary.

jpp · Accepted Answer · 2018-08-11 14:55:52Z

5

You can use np.vectorize with dict.get:

d = np.vectorize(mydict.get)

res = d(input_pattern)

print(res)

array([[1, 6],
       [2, 5],
       [3, 4]])

answered Aug 11, 2018 at 14:55

jpp

166k37 gold badges301 silver badges362 bronze badges

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

1 Comment

tachyon Over a year ago

This is amazing. Thank you.

Tai · Accepted Answer · 2018-08-11 18:44:34Z

0

the problem is there is no way to know the shape of the array beforehand.

We can try to use flat + list comprehension + reshape

np.array([mydict[v] for v in input_pattern.flat]).reshape(input_pattern.shape)

array([[1, 6],
       [2, 5],
       [3, 4]])

flat makes an 1-dimensional iterator out of the array and we later on use reshape to recover the shape of input_pattern.

edited Aug 11, 2018 at 18:44

answered Aug 11, 2018 at 18:37

Tai

8,0643 gold badges31 silver badges50 bronze badges

Collectives™ on Stack Overflow

Constructing a ndarray of values from another ndarray containing dictionary keys

2 Answers 2

1 Comment

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related