A Study on Ensemble Classification Method Based on Accuracy and Confusion Entropy

This paper proposes an ensemble classification method integrating accuracy and confusion entropy to address the limitations of existing ensemble learning approaches that overly rely on single evaluation metrics, particularly in scenarios involving class imbalance and high-dimensional data. The method employs a dynamic weighting mechanism to balance the importance of accuracy and confusion entropy, while Bayesian optimization is used to determine optimal parameters, enabling the effective selection and integration of classifiers. Experimental results demonstrate that the proposed method outperforms traditional approaches on multiple UCI datasets in terms of accuracy,F1 score, AUC, and confusion entropy, showcasing exceptional robustness and adaptability in complex tasks.