In XGBoost the next tree is trained on the residual of previous tree as independent variable or the original independent variable itself?

xgboost

#1

In XGBoost the next tree is trained on the residual of previous tree as independent variable or the original independent variable itself?


#2

Hi @aryan_singh1993,

Each subsequent tree will train on the original independent variables, while considering the residuals as the dependent variable.

Refer the GBM topic in this article -