0
$\begingroup$

XGBoost allows the user to tune both parameters, max depth and min_child_weight. Since both hyperparameters control the depth of trees, it seems that you would only really need to tune one. However, all the blogs/tutorials I see online tune BOTH simultaneously, and I don't understand why.

Is there any reason to tune both?

$\endgroup$
2
  • $\begingroup$ Many more parameters of XGBoost reduce the tree size, and they all do it in a different way. In reality, you don't need to find optimal values for all these parameters simultaneously as they go hand-in-hand and different choices lead to comparable models. $\endgroup$
    – Michael M
    Jan 19, 2022 at 20:10
  • $\begingroup$ @MichaelM But so, are you saying that there is no point in optimizing both, Max Depth and MCW, as they accomplish the same thing? $\endgroup$ Jan 19, 2022 at 21:46

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.