I understand how the cover is calculated in XGBoost, the sum hessian at that node. For the root node of tree 1 for binary logistic, it becomes n(.5)(1-.5) with base score as 0.5. The cover at root node for tree 1 is sum(p(1-p)) where p is the predicted probabilities from the first tree.
What I need is the exact number of observations at each node. Cover "is a metric to measure the number of observations affected by the split"; I need to translate this to the exact number for any given node.
Also, min_child_weight says what is the minimum sum of instance weight needed in a child. So min_child_weight is the minimum 'Cover' in the tree? Correct me if I am wrong.
Any help is much appreciated. Thanks.
