-
Notifications
You must be signed in to change notification settings - Fork 3
Open
Description
If you use a random state to split your test and train data then the model it leads to future look ahead bias because the model already knows what will happen. I havent yet, but if you try the model without randomizing the test set form the dataset it would preform significantly worse.
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
No labels