I know this is explained in the documentation, but even after reading it and trying I am not able to understand it intuitively. A simple explanation would be welcome. Also while modelling how can I decide its value.
$\begingroup$
$\endgroup$
1
-
$\begingroup$ You need to rephrase this to show us what it says and why you do not understand it as (a) we may not read R (b) we may not have the documentation to hand (c) questions purely about how to use R without a reproducible example are off-topic here. $\endgroup$– mdeweyCommented Mar 18, 2017 at 14:26
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
Intuitively, this is the minimum number of samples that a node can represent in order to be split further. If there are fewer than min_child_weight samples at that node, the node becomes a leaf and is no longer split. This can help reduce the model complexity and prevent overfitting.
To select an appropriate value, you can use cross-validation.
