In an XGBoost Regressor, you try to split the sample at node t using feature X(k). The residual array on the left node is [-120, 20]. The residual array on the right node is [-20, 110, 30]. Given the regularization parameter (lambda) = 0 and the minimum loss reduction parameter (gamma) = 10,000. What is the information gain of splitting sample using X(k) at node t?

Answer rating: 100% (QA)

The information gain of splitting the sample using Xk at node t is 6226666666666668 How to calculate ... View the full answer