Selecting random values from dictionary

Question

Let's say I have this dictionary:

dict = {'a': 100, 'b': 5, 'c': 150, 'd': 60};

I get the key which has greatest value with this code:

most_similar = max(dic.iteritems(), key=operator.itemgetter(1))[0]

it returns 'c'

But I want to select a random key from top 3 greatest values. According to this dictionary top 3 are:

c
a
d

It should randomly select a key from them. How can I do that?

So, you want a single random selection from a dictionary? Or a single random selection from a preexisting `list` of the top 3 keys? Or you want to find the top 3 keys, and then select a single random one from them? — TigerhawkT3, Jan 10 '16 at 15:40
I hate to be a downer, but this question is easily answered by combining the answers from [How do I randomly select an item from a list using Python?](http://stackoverflow.com/q/306400/1394393) and [5 maximum values in a python dictionary](http://stackoverflow.com/q/7197315/1394393). I've downvoted as a result, since that constitutes poor research effort in my book. Could you explain how this question distinguishes itself from just slapping those two together? — jpmc26, Jan 11 '16 at 03:02

thefourtheye · Accepted Answer · 2016-01-10T15:47:01.547

13

If you want to find the top 3 keys and then get one of the keys randomly, then I would recommend using random.choice and collections.Counter, like this

>>> d = {'a': 100, 'b': 5, 'c': 150, 'd': 60}
>>> from collections import Counter
>>> from random import choice
>>> choice(Counter(d).most_common(3))[0]
'c'

Counter(d).most_common(3) will get the top three values from the dictionary based on the values of the dictionary object passed to it and then we randomly pick one of the returned values and return only the key from it.

edited Jan 10 '16 at 15:47

answered Jan 10 '16 at 15:41

thefourtheye

233,700
52
457
497

regarding your earlier comment on my answer: does `most_common` not have to do some sorting as well, such that the complexity is O(nlog(n)) for both solutions? – timgeb Jan 10 '16 at 15:49
What about [TigerhawkT3](http://stackoverflow.com/questions/34707280/selecting-random-values-from-dictionary/34707349#34707349)'s answer ? – Iron Fist Jan 10 '16 at 15:52
@IronFist That also has the same O(nlogn) complexity. – thefourtheye Jan 10 '16 at 15:53
2

For picking just 3 elements `most_common` should be closer to linear time http://stackoverflow.com/questions/29240807/python-collections-counter-most-common-complexity – Akavall Jan 10 '16 at 16:20

score 7 · Answer 2 · answered Jan 10 '16 at 15:42

Get the keys with the three largest values.

>>> import heapq
>>> d = {'a': 100, 'b': 5, 'c': 150, 'd': 60}
>>> largest = heapq.nlargest(3, d, key=d.__getitem__)
>>> largest
['c', 'a', 'd']

Then select one of them randomly:

>>> import random
>>> random.choice(largest)
'c'

score 4 · Answer 3 · edited Jan 10 '16 at 16:29

4

Sort the dictionary by descending value, get the first three objects from the resulting list, then use random.choice:

>>> import random
>>> d = {'a': 100, 'b': 5, 'c': 150, 'd': 60}
>>> random.choice(sorted(d, reverse=True, key=d.get)[:3])
'c'

And don't call it dict or you'll mask the built-in.

edited Jan 10 '16 at 16:29

timgeb

76,762
20
123
145

answered Jan 10 '16 at 15:42

TigerhawkT3

48,464
6
60
97

Selecting random values from dictionary

3 Answers3