Python Get Random Sample of List of Tuples

Question

I have a list of over million tuples and want a list of 100,000 tuples randomly sampled from that original list without replacement. I have read:

but the solution provided (using random.sample while casting list to a set) does not work for tuples.

I am using Python 2.7

Can you explain how your data is stored in more detail? The example given works fine for me for lists, tuples, and lists of tuples. — Loocid, Jun 02 '15 at 23:33
The problem comes from casting to a set, as tuples cannot be hashed. The data I am trying to sample without replacement is a list of tuples. — Marjoram, Jun 02 '15 at 23:34
Don't cast it to a set then. `random.sample([(1,2),(3,4),(5,6),(1,41)],2)` should work fine. — Loocid, Jun 02 '15 at 23:36

score 3 · Answer 1 · answered Jun 02 '15 at 23:32

3

Not sure if I fully understand your question, but random.sample can accept both lists and tuples:

random.sample([1,2,3,4,5],2)
Out[620]: [2, 5]

random.sample((1,2,3,4,5),2)
Out[621]: [4, 1]

answered Jun 02 '15 at 23:32

maxymoo

I though random.sample sampled with replacement and that is why, in the link I provided in the question, they cast the list to a set – Marjoram Jun 02 '15 at 23:35
No it's sampling without replacement. – maxymoo Jun 02 '15 at 23:37
1

@Marjoram: No, they call `set` because the question is specifically about sampling from sets, and before the introduction of set literals, `set([a, b, c])` was the way to create a set of `a`, `b`, and `c`. – user2357112 Jun 02 '15 at 23:37

Tommi Kerola · Accepted Answer · 2015-06-03T00:22:49.090

2

You can sample your list regardless of if the elements are hashable or not:

data = create_dataset() # List of data to be sampled

n_samples = 100000
samples = random.sample(data,n_samples)

edited Jun 03 '15 at 00:22

answered Jun 03 '15 at 00:02

Tommi Kerola

2 Answers2