为解决非平衡数据分类中的正样本分类精度不高的瓶颈问题,提出了一种异构分类器融合环境下的非平衡数据分类模型。该模型基于差异采样率的重采样算法和改进的Adaboost算法,融合了SVM和C5.0两种基分类器;基于知识融合机制,采用了独特的分类器选择策略、分类器集成方法、分类决策方案。仿真实验结果表明,SCECM模型分类性能稳定,在非平衡数据集上具有良好的分类性能。
To solve the precision bottleneck of positive sample classification presents the SCECM, a new kind of SVM-C5. 0 ensemble classifier classifier fusion environment. SCECM adopts a differentiated sampling in imbalanced dataset classifying, this paper model which works under the heterogeneous rate algorithm proposed in this study and the improved Adaboost algorithm, takes the two base classifiers of SVM and C5.0, and ttses the unique classifier selection strategy, novel classifier integration approach and original classification decision-making method. Th tion results prove that the SCECM shows the stable and perfect performance when it is applied to imbalan e simula- ced datasets.