Enhanced DNS Cybersecurity: Optimization of XGBoost Using Automated Hyperparameter Tuning via Optuna

This paper presents an application of using Optuna for hyperparameter optimization in a context of improving the performance of an XGBoost model. The starting point from the baseline XGBoost model was an accuracy of around 88.89%. With Optuna-assisted automated and efficient search of the optimal combinations of some hyperparameters-like learning rate, maximum depth, no. of trees, subsampling, column subsampling, min_child_weight-the final accuracy improved considerably to around 93%. Of course, this big improvement underscores how much valuable hyperparameter tuning plays in the improvement of performance in machine learning models. The results establish the ability of Optuna to efficiently find the optimal parameter settings, leading to improved model generalization and predictive accuracy. Beyond improving the robustness of the models, the fine-tuning process enables the construction of a more solid foundation to deploy machine learning models in real-world applications where the high accuracy is obligatory. This research therefore adds to an expanding literature on automatic hyperpa-rameter optimization and the value of tools such as Optuna in optimizing complex models such as XGBoost.