Random-state in Data Science

Question

I know that it's necessary to set Random-state in order to get the same accuracy during a train-test but ask what's the use for, leaving Random-state unset

Answer 1

Usually you only leave it set during model tinkering. If you have done your job well, after finishing your model you can try re-training it with unset random states - if you get similar results, that's good. If not - the model will most likely not generalize well.

Answer 2

Kuba Siekierzyński Nice. Thank you so much