How to obtain reproducible runs in python?

Asked Dec 22 '22 at 10:04

Active Dec 22 '22 at 10:04

Viewed 23 times

I want to use random.seed() so that my runs will be reproducible. I tried the following code to test if using a set seed value works, however it does not provide me with the same output after every run:

import numpy
import random

order = [1,2,3,4,5]
random.seed(1)
random_num = list(numpy.random.permutation(order))

print(random_num)

Now what I want is that when I run the code, it would give me the same output every time. However, this is not what happens. Now I get [1, 3, 2, 4, 5], but the second run gives me [3, 1, 4, 2, 5], which is not the same.

How should I (can I) use the seed value to get reproducible runs?

asked Dec 22 '22 at 10:04

Marloes

Please see https://numpy.org/doc/stable/reference/random/generator.html – medium-dimensional Dec 22 '22 at 10:06
4

`random` and `numpy.random` are entirely different and unrelated beasts! – AKX Dec 22 '22 at 10:06
1

Try `np.random.seed(1)` – Guy Dec 22 '22 at 10:07
2

Just for additional context, `random` is from the Python standard library. `numpy` is an external library. You are seeding different random number generators. :-) – kva1966 Dec 22 '22 at 10:08
You can do the same without numpy and see the effect of the seed: `random.shuffle(order)` – Tomerikoo Dec 22 '22 at 10:10

How to obtain reproducible runs in python?

0 Answers0