0
$\begingroup$

I'm aware that Random Forests are extremely sensitive to overfitting so I'd like to be cautious and make sure I'm checking any outlet possible to avoid it.

I've already split my data into training and test sets but I'm not sure what else to do.

Any information would be much appreciated.

$\endgroup$
1
$\begingroup$

Random forests are generally considered to be relatively robust to overfitting. You want to tune the parameter that controls the number of variables considered for each split. The general rule of thumb is to use p/3, where p is the number of predictor variables, so you might start your search around there. A good algorithm will incorporate bagging and you can find the parameter that gives you the lowest error. You want to grow your forest as large as is useful for minimizing error (further trees don't damage your fit, but just require more computation).

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.