一种不平衡数据流集成分类模型

An Ensemble Classifier Framework for Mining Imbalanced Data Streams

针对不平衡数据流的分类问题,结合基于权重的集成分类器与抽样技术,本文提出了一种处理不平衡数据流集成分类器模型.理论分析与实验验证表明,该集成分类器具有更低的计算复杂度,更能适应存在概念漂移的不平衡数据流挖掘分类,其整体分类性能优于基于权重的集成分类器模型,能明显提升少数类的分类精度.

Many real world data streams mining applications involve learning from imbalanced data streams,where such applications expect to have a higher predictive accuracy over the minority class,however most classification model assume relatively balanced data streams,they cannot handle imbalanced distribution.In this paper,we propose a novel ensemble classifier framework（IMDWE） for mining concept-drifting data streams with imbalanced distribution by using weighted ensemble classifier framework sampling technique including over-sampling and under-sampling.Our empirical study shows that the IMDWE is superior and have improves both the efficiency in learning the model and the accuracy in performing classification over the minority class.