Notes. (1) Learning Rate: Controls step size during boosting, thus determining the contribution of each tree; smaller values require more trees but can improve model generalisation. (2) Max Depth: Maximum tree depth limiting model complexity and preventing overfitting; deeper trees capture more intricate patterns but risk memorisation. (3) Number of Estimators: Total number of trees constructed in the ensemble; more trees can improve predictive performance but also increase computational complexity. (4) Subsample: Fraction of training data used in each tree; this introduces randomness and potentially reduces overfitting; values < 1.0 create stochastic gradient boosting. (5) Colsample by Tree: Proportion of features randomly selected when constructing each tree; this promotes feature diversity and reduces the correlation between trees.