基于随机秩次k近邻规则的不平衡数据分类算法

An Ensemble Imbalanced Data Classification Algorithm Based on Random k-Rank Nearest Neighbor Rules

针对不平衡数据分类问题,为提高二分类任务中少数类样本分类准确率低的问题,本文提出一种随机秩次k近邻集成学习算法——REKRNN。该方法将秩次k近邻算法应用于Bagging集成学习框架中,同时采用混合重采样和随机子空间法平衡训练集,增加基学习器差异性。仿真实验证明,该算法在处理不平衡数据分类任务时性能良好。

In this article, a random ensemble k-RNN algorithm called REKRNN is proposed to deal with the imbalanced data classification. The algorithm incorporates the k-rank nearest neighbor classifier into the frame of Bagging algorithm. At the same time, resampling techniques and random feature method are applied to deal with the imbalanced issue. We observe that the proposed method per-formed remarkably well on different imbalanced dataset. The random ensemble k-RNN algorithm can be considered as a promising tool for imbalanced classification.