Evaluating XGBoost with tidymodels

Cross-validation

Cross-validation is the best practice for evaluating the quality of machine learning models. XGBoost ensembles are no exception to this rule. The following code performs these processes:

1. Prepares the Titanic training data.

2. Specifies the XGBoost classification ensemble with hyperparameter tuning.

3. Creates a tuning grid.

4. Performs five-fold cross-validation for each row of the tuning grid.

5. Outputs the best ten combinations of hyperparameters in terms of accuracy, sensitivity, and specificity metrics.