Python sample without replacement and change population

Question

If you have a list of 100 values, which you want to subset into 3 in the ratio 2:1:1, what's the easiest way to do this in Python?

My current solution is to take a sample of the indices for each subset then remove these values from the original list, i.e.

my_list = [....]
num_A = 50
subset_A = []

num_B = 25
subset_B = []

num_C = 25
subset_C = []

a_indices = random.sample(xrange(len(my_list)), num_A)
for i in sorted(a_indices, reverse=True):  # Otherwise can get index out of range
    subset_A.append(my_list.pop(i))

b_indices = random.sample(xrange(len(my_list)), num_B)
for i in sorted(b_indices, reverse=True):  # Otherwise can get index out of range
    subset_B.append(my_list.pop(i))

subset_C = my_list[:]
assert len(subset_C) == num_C

However I'm sure there's a much more elegant solution than this.

If the length of my_list is 100, and a_indices starts off with [21, 5, 99...] then when it comes to pop the 99th index out of a list with now 98 elements it will be out of range. — Stuart Lacy, Oct 15 '14 at 10:13

score 5 · Accepted Answer · answered Oct 15 '14 at 10:10

5

There's a much easier way. You can just shuffle the array and take parts.

xs = [...]
random.shuffle(xs)
print(xs[:50], xs[50:75], xs[75:])

answered Oct 15 '14 at 10:10

simonzack

19,729
13
73
118

Exactly what I was looking for, didn't come across my mind to use shuffle. – Stuart Lacy Oct 15 '14 at 10:14

Python sample without replacement and change population

1 Answers1