How do I randomize a pandas column in the same order every time I run the code in Python?

Question

This is my code:

random_idx = np.random.permutation(len(cIds))
train_Ids = cIds[random_idx[:train_size]]

Now, I want the list to be randomized in the same order every time I run this line of code.

Note: I do not want to save random_idx variable in a text file, and fetch the same list.

Possible duplicate of [random.seed(): What does it do?](https://stackoverflow.com/questions/22639587/random-seed-what-does-it-do) and [this comment](https://stackoverflow.com/questions/22639587/random-seed-what-does-it-do#comment74255398_22639752) is particularly useful. — pault, Dec 22 '18 at 04:42
You need to use a `seed` as is demonstrated in [this answer](https://stackoverflow.com/questions/47742622/np-random-permutation-with-seed). — jason m, Dec 22 '18 at 04:44
What is the range of seed value? like 0-100? -100 to 100? infinite ot infinite? Can it be negative? Fraction? — John Doe, Dec 22 '18 at 04:46

Vlad Bezden · Accepted Answer · 2018-12-22T05:19:23.127

6

You can use seed in order to tell numpy to generate the same random numbers:

np.random.seed(seed=1234)
random_idx = np.random.permutation(len(cIds))

same as:

np.random.seed(1234)
random_idx = np.random.permutation(len(cIds))

Or

random_idx = np.random.RandomState(seed=1234).permutation(len(cIds)

seed: Must be convertible to 32 bit unsigned integers`

edited Dec 22 '18 at 05:19

answered Dec 22 '18 at 05:10

Vlad Bezden

score 1 · Answer 2 · answered Dec 22 '18 at 06:58

1

Or you can do in in pandas style:

cIds.sample(n=train_size, 
            replace=False, 
            random_state=your_favorite_number_here)

answered Dec 22 '18 at 06:58

Quang Hoang

2 Answers2