random.seed() does not work with np.random.choice()

Question

So i'm trying to generate a list of numbers with desired probability; the problem is that random.seed() does not work in this case.

M_NumDependent = [] for i in range(61729): random.seed(2020) n = np.random.choice(np.arange(0, 4), p=[0.44, 0.21, 0.23, 0.12]) M_NumDependent.append(n) print(M_NumDependent)

the desired output should be the same if the random.seed() works, but the output is different everytime i run it. Does anyone know if there's a function does the similar job of seed() for np.random.choice()?

willeM_ Van Onsem · Accepted Answer · 2024-02-13 14:30:38Z

numpy uses its own pseudo random generator. You can seed the Numpy random generator with np.random.seed(…) ^[numpy-doc]:

np.random.seed(2020)

For example:

>>> np.random.seed(2020) >>> np.random.choice(np.arange(0, 4), p=[0.44, 0.21, 0.23, 0.12]) 3 >>> np.random.seed(2020) >>> np.random.choice(np.arange(0, 4), p=[0.44, 0.21, 0.23, 0.12]) 3 >>> np.random.seed(2020) >>> np.random.choice(np.arange(0, 4), p=[0.44, 0.21, 0.23, 0.12]) 3 >>> np.random.choice(np.arange(0, 4), p=[0.44, 0.21, 0.23, 0.12]) 2

As you can see we each time pick 3 whereas if we do not seed the random generator, 2 is the next item after 3.

Giorgos Myrianthous · Accepted Answer · 2020-04-16 18:16:40Z

You are accidentally setting random.random.seed() instead of numpy.random.seed().

Instead of

random.seed(2020)

use

import numpy as np np.random.seed(2020)

and your results will always be reproducible.

I think you mean random.seed() instead of random.random.seed() ?

Collectives™ on Stack Overflow

random.seed() does not work with np.random.choice()

2 Answers 2

Comments

1 Comment

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

1 Comment

Related