What is the use of numpy.random.seed() Does it make any difference?

Question

I have a dataset named "admissions".

I am trying to carry out holdout validation on a simple dataset. In order to carry out permutation on the index of the dataset, I use the following command:

import numpy as np
np.random.permutation(admissions.index)

Do I need to use np.random.seed() before the permutation? If so, then why and what does the number in np.random.seed(number)represent?

Solution 1

You don't need to initialize the seed before the random permutation, because this is already set for you. According to the documentation of RandomState:

Parameters:
seed : {None, int, array_like}, optional Random seed initializing the pseudo-random number generator. Can be an integer, an array (or other sequence) of integers of any length, or None (the default). If seed is None, then RandomState will try to read data from /dev/urandom (or the Windows analogue) if available or seed from the clock otherwise.

The concept of seed is relevant for the generation of random numbers. You can read more about it here.

To integrate this answer with a comment (from JohnColeman) to your question, I want to mention this example:

>>> numpy.random.seed(0)
>>> numpy.random.permutation(4)
array([2, 3, 1, 0])
>>> numpy.random.seed(0)
>>> numpy.random.permutation(4)
array([2, 3, 1, 0])

Solution 2

Note that np.random.seed is deprecated and only kept around for backwards-compatibility. That's because re-seeding an existing random-number generator (RNG) is bad practice. If you need to seed (e.g., to make computations reproducible for tests), create a new RNG:

import numpy as np


rng = np.random.default_rng(seed=0)
out = rng.random(5)