Why a train/test split?
What is the point of making a train/test split for binary classification problems?
Answer the question
To make the problem harder for the model by reducing the dataset size.
To evaluate your models out-of-sample, on new data.
To reduce the dataset size, so your models fit faster.
There is no real reason; it is no different than evaluating your models in-sample.
Take Hint (-15xp)