Enhancing Unbalanced Data Classification with Cross-Validation and Extreme Gradient Boosting: A Comprehensive Analysis

As a novel and efficient ensemble learning algorithm, XGBoost has been widely applied due to its multiple advantages, but classification effect in cases of data imbalance is often not ideal. Aiming at this problem, efforts were made optimize the Cross Validation algorithm. The main idea combine cross validation on unbalanced for processing, then get final model based through training. At same time, optimal parameters are searched adjusted automatically optimization algorithms realize more accurate predictions. In testing phase, area under curve (AUC) used as an evaluation indicator compare analyze performance various sampling methods algorithm models. results analysis using AUC expected verify feasibility effectiveness proposed