Predicting Cancer Classification Based on the Improved XGBoost Algorithms

Objective: This study aims to explore the optimization of XGBoost algorithm parameters based on heuristic algorithms, with the goal of improving the classification accuracy of the algorithm.Methods: Leveraging the principles of bionics, we employed genetic algorithm (GA), dung beetle optimizer (DBO), and grey wolf optimizer (GWO) to optimize the parameters of XGBoost. Using the Wisconsin breast cancer binary classification dataset as a basis, experiments were conducted with 70% and 80% of the training data to validate the effectiveness of the proposed algorithms.Results: Under both the 70% and 80% training set scenarios, DBO_XGBoost and GWO _ XGBoost demonstrated superior performance compared to GA XGBoost in most evaluation metrics. DBO XGBoost _ _ exhibited robust performance in both settings, while GWO_XGBoost achieved optimal performance with the 80% training set. Conclusion: The DBO _ XGBoost algorithm exhibited excellent performance under both training set configurations, demonstrating its powerful optimization capabilities and broad applicability. The utilization of heuristic optimization algorithms for parameter optimization of XGBoost significantly enhances model performance.