Splitting with hashlib

Good time a day ageron, you have a very nice book.

Anyway I have one question from Chapter 2. Which of algorithm given in book is more faster:

**- 1) def split_train_test(data, test_ratio)
- 2) def test_set_check(identifier, test_ratio, id_column, hash=hashlib.md5)
- 3) train_test_split from sklearn**

?

Which function is more save memory?

I would be gratefull if you give comments about this. Thanks for your attention