While tuning a tree-based ensemble model with AUC as the objective, which visualization helps decide how to adjust a hyperparameter range (for example, maximum tree depth) to reduce training time and cost?

A scatter plot showing the relationship between maximum tree depth and the objective metric (AUC) directly illustrates how different depth values impact model performance. This visualization helps identify optimal depth ranges and understand diminishing returns, allowing you to narrow down the search space for hyperparameter tuning. This reduces the number of trials needed, thereby saving training time and cost. Incorrect options: A histogram of an input feature's distribution is useful for understanding data characteristics but doesn't directly inform hyperparameter tuning for tree depth. A t-SNE plot helps visualize high-dimensional data clusters but isn't directly used for optimizing a specific hyperparameter like tree depth. A scatter plot displaying the objective metric's value across training iterations shows convergence during a single training run, not the impact of varying a hyperparameter across multiple runs.