Please help me regarding Decide how to apply random_state in dataset and when to apply?

quetion1- when to choose random_state in train_test_split and when to not
question- 2 how we decide what to put in random state (None or integer) and how they gonna affect the training dataset.
questioin 3 - also if we choosing randaom_state = any integer , how to know which integer value i should put to get good trainimg result

Q1: always specify a random_state.
Q2: always an integer
Q3: there is no rule of thumb as to which integer will give you the best luck. However, it is a good practice to always specify the random_state such that your results are reproducible. Reproducibility is important for experimentation and reproducing results.

Here is a DLS lecture on what dev/test sets are good

Raymond